In discussions about AGI, no word comes up more often than "consciousness." Some argue that a machine must "achieve consciousness" to truly think — that without subjective experience, without qualia, the sense of "what it is like" to be something, intelligence remains mere imitation. This position draws on a half-century tradition from John Searle, 1980 to David Chalmers, 1995 on the necessity of subjective experience, and it is intuitively compelling.

There is another word — less dramatic but more important for our purposes: "reflection." The simple ability of a system to evaluate its own output, compare it against a criterion, and correct itself. A functional loop possessed by any thermostat, by a DNA polymerase molecule, and — in distributed form — by an ant colony. None of these systems is conscious, yet all of them are reflective. Besides reflection, consciousness includes qualities such as:

• intentionality — the directedness of attention toward a specific object
• integrativity — the capacity to unify disparate signals into a coherent whole
• temporality — the ability to retain the past and anticipate the future over extended durations
• sense of agency — the feeling of authoring one's own actions, the mechanism that distinguishes "I do" from "it happens to me"

Consciousness typically includes reflection as just one of its mechanisms. But reflection is possible outside consciousness, as a pure mechanism of interaction between an object and its environment. Moreover, unlike consciousness, reflection appears at virtually every level of material organization where the system is sufficiently complex. And most of the practical properties we attribute to a "conscious" AGI — planning, self-correction, awareness of one's own limitations — are in fact properties of reflection.

Reflection — a system's ability to evaluate its own output and correct it — requires neither consciousness, nor intention, nor subjective experience. It arises at every level of material organization, from quantum particles to scientific communities. And it is precisely this component of consciousness that is being leveraged today in attempts to build AGI.

To see why, we will trace reflection from the bottom up: from abstract information through inanimate and living matter to collective intelligence.

Cybernetics: The Universal Schema of Reflection

In 1948, Norbert Wiener proposed in his paper Norbert Wiener, 1948 that any self-correcting system consists of at least three elements linked by a feedback loop:

1. Receptor / Sensor — a sense organ or detector that captures information about the state of the external environment.
2. Comparator — an element that compares the sensor's reading with a target value and computes the difference.
3. Effector — an actuator or part of the system that performs an action (e.g., a motor, a muscle, or any mechanism that does work).

Wiener called the discipline that studies such systems cybernetics — the science of control and communication in living organisms and machines. The word comes from the Greek κυβερνήτης ("helmsman"): like a ship's helmsman, the system continuously compares the current course with the goal and corrects its trajectory (Fig. 1).

For Wiener, control alone was not enough — communication mattered just as much: the system must not merely act but transmit the error signal between its components. At the same time, feedback is useful only as long as it does not drive the system into oscillations and instability. In other words, reflection emerges as the result of a loop that keeps the system in a stable operating regime. This is what creates the impression of a "smart system" — one that adapts to external conditions.

Wiener's schema — sensor, comparator, effector — is a universal language for describing any form of self-correction. Now let us trace how this pattern manifests at each level of material organization, starting with the most abstract.

Level 1. The Abstract: Self-Correction of Pure Information

The purest example of reflection is one involving neither matter nor life — just information defending itself against chaos. Hamming codes (Fig. 2) are a practical example of "digital homeostasis." Special parity bits are added to useful data, each monitoring a specific group of bits. If an external disturbance — say, interference in a cable — flips a single data bit (0 → 1), the system instantly "senses" the problem:

• Stress (Disturbance): External noise inverts one bit in the sequence. In the diagram, this is the transformation of a "one" into a red zero at position 4. The system transitions from a stable state to one containing an "information defect."
• Response (Comparator): The parity bits, embedded in the data structure, "detect" the anomaly. A mathematical algorithm performs a parity check: the checksums no longer match, producing a nonzero error "syndrome."
• Correction (Effector): Based on the syndrome, the system computes the exact coordinate of the failure. At the intersection of the parity bits' responsibility zones, an inverse operation is triggered: inversion 0 → 1. The loop closes, restoring the data block to its original stable state.

Wiener's pattern works in the realm of abstractions: state → disturbance → correction. Hamming codes are the simplest case, correcting a single bit. In real information systems, the same principle scales into multi-layered protection: Reed–Solomon codes restore entire data blocks on damaged CDs and DVDs; the TCP protocol retransmits lost packets using checksums and timeouts; RAID-6 allows a disk array to survive the simultaneous failure of two drives. In every case, information protects itself through redundancy — a reflective loop embedded in the structure of the data.

Level 2. Inanimate Matter: Reflection Without a Designer

Quantum reflection: a single particle. At the quantum level, feedback-like mechanisms appear even in the behavior of individual particles. Here, Wiener's "receptor," "comparator," and "effector" are fused with the very fabric of reality. The Pauli exclusion principle, for instance, determines the particular structure of matter. Every electron in an atom is described by a set of quantum numbers — a kind of "address" specifying its orbit, energy, and spin direction. The Pauli principle forbids two electrons from sharing the same address.

Suppose two electrons are pushed toward the same state with identical quantum numbers (Fig. 3). The moment an external force — say, the colossal pressure inside a dying star — tries to "squeeze" a particle into an already occupied state, the system instantly responds with a counterforce. The system's wave function "senses" the violation of a fundamental symmetry and generates what is known as degeneracy pressure. The outcome of the process is directed at compensating the disturbance (the attempt at compression), which prevents the collapse of atoms and allows solid matter to exist. The system corrects its state because any other behavior is forbidden by the geometry of space itself. And every electron obeys this principle.

Strictly speaking, there is no separate "sensor" or "comparator" from Wiener's schema here — there is a fundamental symmetry prohibition that makes certain states impossible. This is more of a metaphorical extension of the concept of reflection: correction is built into the physics itself rather than arising as a superstructure. Nevertheless, the result is the same — the system returns to an allowed state after a disturbance, and it is precisely this functional pattern that links the quantum level to the other rungs of the ladder.

Statistical reflection: an ensemble of particles. When the number of particles grows large, a new level of reflection emerges — a statistical one. A striking manifestation of statistical reflection at the inorganic level is Le Chatelier's principle: a chemical system at equilibrium, when subjected to a disturbance, shifts so as to partially compensate for it. A vivid example of such "natural reflection" is the classic cobalt chloride experiment. Suppose a solution contains two ions in equilibrium (Fig. 4):

The outcome of the process (release or absorption of heat) is directed at compensating the cause (the external temperature change), and the system does this on its own, without sensors or wires. In essence, this is a property of thermodynamics — a consequence of Gibbs free energy minimization. Note that for a single molecule, the concepts of "equilibrium" or "pressure" do not exist. Reflection emerges only when the number of particles is sufficiently large.

Thus, even at the level of inanimate nature, self-correction mechanisms exist — without cells, without neurons, without intention. But such systems have only a single loop: they respond to a disturbance without "remembering" previous ones.

Level 3. The Cell: A Hierarchy of Loops

The cardinal distinction of biological systems is the multi-level nature of their reflection. Even a single cell is not one feedback loop but an entire hierarchy of nested loops, where each successive one activates only when the previous one has failed.

DNA polymerase, for example, works like a typewriter with a built-in eraser: if it places the wrong "letter," it physically cannot proceed until it "erases" the error and replaces the nucleotide with the correct one (Fig. 5). If this instant self-check fails, the next tier of quality control kicks in: specialized editor proteins re-examine the finished text. At the whole-cell level, chaperones attempt to refold misshapen proteins. Finally, in a critical situation where accumulated errors threaten the organism, the system makes a final reflective decision — triggering apoptosis (self-destruction) to prevent the buildup of chaos.

The fundamental difference between living and inanimate matter is the emergence of memory and composite hierarchy. Chemical equilibrium does not remember past disturbances; a cell does — and builds defenses accordingly. Reflection ceased to be a one-off reaction and became a multi-layered process. Yet the cell is still unable to evaluate the quality of its own evaluation. That leap occurs at the level of the organism.

Level 4. The Organism: From Body to "Knowing What You Don't Know"

Proprioception is a basic example of reflection at the organism level. The brain predicts the sensory outcome of a movement, compares the prediction with the actual result, and immediately makes corrections. Imagine you are about to lift a large box that you believe is packed full of books.

1. Prediction: The brain calculates the required effort in advance and commands the muscles to brace for a 15 kg pull.
2. Reality: The box turns out to be empty. The moment you begin the motion, your arm shoots upward far faster than expected.
3. Comparison (prediction error): At this point, the "proprioceptive sensor" (receptors in muscles and tendons) registers a massive mismatch between plan and reality.
4. Correction: The brain instantly, within fractions of a second, "reflects" on this error and sends a braking signal so you don't smack yourself in the face with the box.

This cycle of "Prediction — Comparison — Correction" runs thousands of times per second: when you walk on uneven ground, type without looking at the keyboard, or try to touch the tip of your nose with your eyes closed. There is no conscious deliberation here — it is a pure cybernetic loop in which movement self-corrects during execution.

Metacognitive reflection: "Do I know what I know?" Proprioception is reflection of the body. But organisms are capable of something more: evaluating not just the external signal, but the reliability of their own processing of that signal. This is a qualitative leap — reflection upon reflection.

In experiments by J. David Smith (Smith et al., 1995) with bottlenose dolphins, the capacity of mammals to monitor their own uncertainty was demonstrated. A dolphin was asked to distinguish between sound tones of different frequencies (Fig. 6). For correctly identifying the higher-frequency source, the dolphin received a reward; for an error, a temporary penalty — a delay before the next trial. The researchers then introduced a third option: the ability to "decline" the trial. It turned out that when the frequency difference became critically small and the task was objectively difficult, the dolphin preferred to press the decline button, avoiding the penalty and moving to the next round. This is how the internal reflective loop works: the animal evaluates not just the external signal, but the reliability of its own brain's processing of that signal, choosing a strategy that minimizes risk.

Analogous results have been obtained in studies of rhesus macaques (Middlebrooks & Sommer, 2012, Neuron; Hampton, 2001, PNAS). Monkeys were given a visual memory task and then had to choose the size of their reward: a guaranteed small prize or a large prize with the risk of total loss if they were wrong. Macaques chose the high-risk option only when their memory gave a clear answer, and switched to the conservative strategy — a bird in the hand — when their internal model flagged gaps in the data. Studies of neural activity showed that at this moment, a specific "confidence prediction error" signal forms in the prefrontal cortex and anterior cingulate cortex of the macaque brain — effectively a biological analogue of a metacognitive comparator.

Deferred reflection. Biological reflection does not cease when consciousness switches off; during sleep, it shifts into a mode of deep systemic optimization. During slow-wave sleep, memory consolidation takes place: the hippocampus begins replaying the day's accumulated experience to the neocortex at an accelerated pace. In essence, this is a reflective "replay" of events to separate the important from the incidental. The brain analyzes the day's prediction errors and updates its internal proprioceptive and cognitive models, freed from the distraction of incoming sensory input. This "audit" allows the system to resolve information conflicts and restructure long-term memory, transforming chaotic experience into ordered knowledge.

In summary, the complex hierarchy of cells within an organism adds new reflective loops on top of the reflection of individual cells: prediction (proprioception), evaluation of one's own confidence (metacognition), and optimization of reflection itself (sleep). None of these properties requires consciousness — the dolphin does not reflect on "what it is like to be a dolphin"; it simply presses the decline button when it is not sure.

Level 5. The Swarm: Distributed Reflection Without a Reflector

In distributed systems, reflection is implemented through external interaction markers of three types: material traces in the environment, coordination signals, and persistent records of collective experience.

Stigmergic reflection. In this framework, coordination arises through the environment — a mechanism known as stigmergy: the trace of one agent's action becomes a guide for others. In ants, these external markers are pheromone trails; the colony's memory is, in a sense, externalized into the environment in active form. An individual ant acts chaotically and possesses no global plan, but its interaction with pheromones creates a feedback loop: the shorter the path to food, the more frequently ants traverse it per unit of time, and the stronger the chemical signal becomes for the rest of the colony.

Pheromone evaporation serves as an "error filter" (discarding suboptimal routes), while accumulation serves as "reinforcement of successful solutions." If an obstacle appears on a path, the system responds to the "stress" with an instant redistribution of traffic: the old trail fades and the detour strengthens automatically, with no centralized command. The colony reflects changes in the landscape, using itself as a computational network.

Collective skepticism and independent verification. In a honeybee colony, the self-correction mechanism reaches a new level of sophistication through the combination of collective signaling and individual experience. When a scout bee returns to the hive and signals a foraging site by performing the "waggle dance," other bees do not simply copy her behavior.

• Individual filter: A recruit bee reads the coordinates from the scout's dance, but before relaying this signal further, she must personally fly to the site and assess the quality of the resource.
• Correction through comparison: If the nectar turns out to be scarce or conditions at the site have changed, the recruit will not dance upon returning. The signal about a poor route simply fades, receiving no confirmation.
• Mathematical robustness: Thanks to this "independent verification," the swarm is protected from cascading errors. The system reflects reality through thousands of individual filters, allowing the colony to select the best food source from dozens available.

Epistemic stigmergy. In human culture, the scientific method performs the same function as pheromone trails or bee dances — it is a protocol of interaction that transforms the chaotic activity of individual researchers into a self-correcting process of knowledge production. Scientific papers are the analogue of chemical pheromones — "markers" in the environment that point other agents toward promising directions. Citation works like pheromone accumulation: the more frequently a result is reproduced and cited, the "stronger" that route becomes in the landscape of knowledge. The specialization of scientists allows the system to process vast datasets in parallel: thousands of labs, like scouts in a swarm, probe different hypotheses simultaneously, requiring no centralized command.

In such swarm systems, there is no unified consciousness. There is distributed reflection without a reflector — and it works. Our consciousness immediately interpolates an agent: "nature decided," "evolution optimizes," "the model understands." This is convenient language but poor ontology: the feedback loop operates without the intention of any subject. In this respect, swarm systems resemble the molecular systems we examined earlier when discussing Le Chatelier's principle.

Up to this point, each new level of complexity has yielded deeper reflection. A particle resists disturbance. A molecular ensemble shifts its equilibrium. A cell proofreads its own work. An organism predicts the consequences of its own actions and evaluates its confidence. A collective optimizes routes and knowledge. One might think the trend continues without limit: keep increasing complexity, and reflection will keep deepening. But this ladder has a ceiling.

The Meta-Level. Recursive Self-Observation and the Limits of Control

Second-order cybernetics: "The observer within." If classical cybernetics (Wiener) studied how a system controls an object, second-order cybernetics (Heinz von Foerster, 1979; Heinz von Foerster, 1991) focuses on systems that include the observer: the system manages not only some correctable parameter but reflects on how it itself observes and makes decisions. The next step follows naturally — if the observer is part of the system, one needs to formally describe what remains stable in recursive self-observation and is recognized as "the same." In von Foerster's framework, this role is played by the concept of eigenforms.

Eigenforms (Heinz von Foerster, 1976) are the "proper forms" of a system that emerge from recursive processes. Infinite application of certain mathematical functions to themselves ($f(f(f(...)))$) stabilizes them at a particular value — the mathematical equivalent of a stable "self." A biological example of such a mechanism is vision: we do not see a "picture" but rather the result of the brain's recursive verification of what it expects to see. The system "reflects" on its expectations until they converge with the signal — this is demonstrated, for instance, by experiments with inverting goggles (George M. Stratton, 1896; George M. Stratton, 1897): when worn continuously, the brain gradually adapts and "flips" the image back.

At the software level, reflection manifests through an algorithm's ability to treat its own code as data. The mathematical embodiment of this principle is quines — programs that, when executed, output their own exact source code. The existence of quines proves that self-reference is not a logical error: any sufficiently complex system can contain within itself a complete set of instructions for assembling itself, much as DNA contains the blueprint for a cell capable of copying that very DNA.

Von Foerster's practical insight is that reflection is not merely a system's "internal commentary" about itself, but rather a practical mechanism for stabilizing invariants through recursive self-checking. Therefore, for complex systems (including LLMs), it is important to evaluate not only the accuracy of an answer, but also which stable forms the system preserves upon repeated checking and how it restructures them when an error occurs.

The computational mechanism: the reflective tower. If eigenforms describe the stable outcome of recursive self-observation, in programming such a mechanism is typically modeled through a hierarchy of interpreters, where each upper level treats the one below not as "external data" but as its own modifiable object. In other words, this is the recursion of an observer over an inner observer: an interpreter can interpret and rewrite the interpretation rules of the previous level. Ascending this "tower" of reflection, a program (for example, the metacircular evaluator in Lisp; see Harold Abelson, Gerald Jay Sussman, Julie Sussman, 1996) can modify its own logic during execution. This is the digital analogue of DNA polymerase's built-in "eraser": the system detects an inefficiency in the base protocol and dynamically corrects it, turning reflection into an instrument of fundamental self-modification.

The practical consequence for development: a mature system should have not only a "working" layer but also a meta-layer of self-observation and self-correction. In engineering terms, this means observability, automated checks, canary releases, rollback, and loops like generate → critique → revise in LLM systems. In other words, a product's robustness is determined not by the absence of errors but by the architecture's ability to detect and fix its own errors before they become systemic.

The logical poison: Gödel's theorems. In 1931, Kurt Gödel mathematically demonstrated the limits of self-reflection. To grasp the essence of Gödel's idea in simplified terms, imagine a "Perfect Logic Textbook" that claims to have answers to all questions. Gödel proved that if this textbook is thick enough to describe itself, one peculiar page will inevitably appear in it.

Imagine we read the following statement on that page:

"This statement cannot be proven by the rules of this Textbook."

Let us try to verify this statement using only the Textbook's own logic:

1. Suppose we manage to prove it. But the statement itself says it cannot be proven. So the Textbook has proven a falsehood. Conclusion: The Textbook is broken (inconsistent).
2. Suppose we cannot prove it. Then it turns out the statement is true! (After all, it claimed exactly that — that it cannot be proven.)

In other words, in any sufficiently powerful formal system of logic, one can construct a statement that asserts its own unprovability. If the system proves it, the system is inconsistent; if it does not, the system is incomplete. To fully "understand" the system, one must step outside it into a meta-system. But there, the same problem repeats.

Gödel's two incompleteness theorems establish a hard epistemological limit for any complex system: complete self-knowledge from within is technically impossible. The first theorem states that in any sufficiently powerful formal system, there will always be true statements that cannot be proven by the system's own means — meaning "truth" is always broader than "provability," and the system will never contain the full totality of knowledge about itself. The second theorem is even more devastating, asserting that a system cannot prove its own consistency while remaining within its own rules. To verify the absence of internal errors and logical dead ends, the system must ascend to a meta-level and examine itself "from outside" — yet that external vantage point itself becomes a new system with the same blind spots (Fig. 9).

Turing demonstrated the computational analogue of Gödel's limitation in the world of programs (Alan M. Turing, 1936/1937). Suppose we have a "perfect predictor" that, for any program, tells us in advance and without error whether it will halt or loop forever. Then one could write a paradox program that does the opposite: if the predictor says "halts," it enters an infinite loop; if it says "loops," it halts immediately. If we run this program on itself, any answer the predictor gives will be wrong. Therefore, a universal and infallible halting test cannot exist in principle.

So reflection has a fundamental limit: a system can correct itself, but it cannot obtain complete knowledge of its own behavior from within alone. In formal logic, this manifests as incompleteness (Gödel); in computability theory, as the undecidability of the halting problem (Turing). But this does not make reflection useless — merely an open, never-ending process.

Implementing Reflection in Large Language Models

In modern large language models, the self-attention mechanism is often mistakenly equated with digital reflection. At each layer of the Transformer architecture, every token does indeed "look back" at the entire preceding context, mathematically weighting the relevance of each word. The model appears to evaluate what has already been written, selecting the most accurate continuation. However, unlike a living cell or a brain, this process lacks the essential attribute of reflection — a closed feedback loop.

New tokens have no effect on previous ones; information flows strictly in one direction, from input layer to output layer. The standard architecture lacks recurrent loops or internal control mechanisms that could compare the result with an expectation and send a corrective signal back into the system before the next word is generated. If DNA polymerase can "erase" an error and a bee can "change its mind" after verification, a transformer simply computes the most probable next step. This is static context analysis, not dynamic self-correction. In this sense, modern LLMs are vast "prediction machines" that simulate consequences but do not control causes.

Decorative post-reflection. Modern machine learning engineering is shifting from "fast thinking" architectures (System 1, per Daniel Kahneman, 2003) toward "slow thinking" systems (System 2) capable of internal delay and verification. If the classic transformer is a straight line, new approaches try to bend it into a loop. One popular approach to overcoming this linearity is the chain-of-thought technique: the model is made to "think aloud." Each newly generated reasoning step enters the context window. Subsequent layers of the network see the model's earlier "thoughts" as part of the input. This creates the simulation of an internal comparator: the system gains the ability to correct course based on its own intermediate conclusions. However, post-reflection is a fragile way to add iterativeness. A 2025 study by Anthropic (Yanda Chen et al., 2025) showed that when using such mechanisms, the neural network tends not so much to reflect as to rationalize — fitting an explanation to a decision already made.

Swarm reflection through human evaluators. Between post-reflection at the inference stage and self-correction at the training stage lies an intermediate loop that directly mirrors the swarm logic from the previous section. In RLHF (Reinforcement Learning from Human Feedback; Long Ouyang et al., 2022), human evaluators play the role of recruit bees: they receive the model's "dance" (a pair of responses), personally assess the quality of each, and vote for the better one. From thousands of such individual assessments, a reward model is constructed, which then guides training. As in a bee colony, collective skepticism is at work: no single evaluator determines the direction alone, and cascading errors are dampened by the independence of judgments. An extension of this idea is RLAIF (RL from AI Feedback), where the role of "recruits" is taken by another model, closing the loop within the machine system. This is the largest-scale reflective loop in modern LLMs: it is slow (update cycles take weeks), but it is what shapes the model's baseline behavior.

Reflection at the training level. If post-reflection in "think aloud" mode often leads to rationalization, the next step is to move self-correction from the inference stage into the training stage itself. This idea is realized by the SCoRe method (Aman Kumar et al., 2024), where the model is trained not only to produce an answer but to correct its own errors across successive attempts. The key difference from the chain-of-thought approach is that corrective behavior becomes part of the model's weights rather than a transient effect of a clever prompt. This shifts the focus from "explain your solution" to "improve your solution": the system must increase the quality of its second attempt relative to the first, rather than simply producing an elegant justification for its initial answer.

However, feedback in such a mechanism is both slow and expensive: first the model makes an attempt, then a correction, then a reward is computed over this chain, and only then are parameters updated. This makes training slower and less stable than conventional methods, and introduces a quality-versus-cost tradeoff at inference time. Additional self-correction iterations yield better answers but increase latency and expense.

Reflection at the architecture level. While training attempts to embed self-correction in the model's weights, the architectural approach changes the very geometry of computation. Instead of a unidirectional pass from input to output, researchers introduce recurrent loops (e.g., in Mostafa Dehghani et al., 2018, Universal Transformers): the model can return to an intermediate state and decide for itself how many iterations a task requires. In such a design, "thinking time" becomes a manageable resource rather than a fixed number of layers. This directly echoes the section on dolphins and macaques: when confidence is low, the system opts not for the "fastest answer" but for a more cautious strategy, increasing verification time or reducing error risk.

The problem is that such architectures align poorly with the economics of modern hardware. Standard transformer training maps well onto GPUs with straightforward parallelization, whereas sequential recurrent steps introduce noticeable speed and cost penalties. In other words, "honest" architectural reflection often proves too expensive at scale.

Hence the interest in hybrid compromises: what the industry needs is not the "perfectly reflective" model but one that is "smart enough and fast enough." Architectures like Mamba (Albert Gu, Tri Dao, 2023) and RWKV (Bo Peng et al., 2023) address a common practical challenge: avoiding the need to recompute the entire preceding text at each step, as a classic transformer does. Instead, the model maintains a compact "working state" and updates it as tokens are read. This allows processing longer inputs within the same budget and lowers inference costs. The downside is that on certain tasks quality may be less consistent, and the ecosystem around these models is still thinner: fewer ready-made tools, proven recipes, and operational practices than for classic transformers.

This is precisely why the industry has massively shifted toward procedural simulation of architectural reflection at the inference stage. The model uses its own intermediate text as temporary working memory, re-reads it, and corrects course. This is a practical compromise: the architecture remains nearly linear, but behavior becomes iterative through additional computational steps.

Conclusion

Consciousness includes reflection, but reflection requires neither consciousness, nor intention, nor subjective experience. A closed feedback loop is sufficient. This is why it works even where there is no "inner observer" in the human sense: amid noise, uncertainty, and complex signals. Thus a ladder of increasing sophistication emerges: from a single-loop response to hierarchies of loops, from reacting to disturbances to prediction, from prediction to evaluating one's own confidence, from individual correction to distributed self-organization.

Against this backdrop, the gap in modern LLMs becomes especially visible. The classic transformer architecture remains linear: the pass goes forward, with no built-in self-checking loop. What looks like "reflection" in commercial models is more often implemented procedurally: the model verbalizes intermediate steps and then reads them back as new input. This is a workable engineering trick, but it is still not architectural reflection — it is its operational simulation.

The practical takeaway, however, is powerful: systems without intention can solve problems no less effectively than "intentional" systems, provided the task is specified cybernetically — through a goal, an error signal, and a corrective loop. And reflection remains the only property traditionally associated with consciousness that we already know how to design, measure, and scale. That is why the current research program is not metaphysical but engineering: not "how to engender experience" but "how to make the loop more robust." The more complex the task, the more nested loops of checks and balances the system requires; no additional "metaphysical layer" is needed for this.

At the end of the 18th century, English lawyer Jeremy Bentham became disillusioned with the state of British law, which was rather confusing at the time. Bentham proposed a principle: to consider "right" any action that, all other things being equal, produces the greatest amount of good for the greatest number of people. This principle is easy to accept at face value, until we ask who determines what constitutes "good" and what to do when the interests of different parties conflict.

AI agents find themselves at the crossroads of ethical obligations without clear instructions. Autonomous agents based on artificial intelligence systems find themselves at the crossroads of ethical dilemmas without clear instructions. Several interests collide in each of their actions: the interests of a specific user, the requirements of developers, social norms, and the internal evaluation system of the agent itself. When an engineer uploads code to Cursor, a doctor consults a medical AI agent about a diagnosis, and a financial analyst uses AI for investment predictions, each of them finds themselves in a unique ethical universe with its own laws.

Without a clear hierarchy of values, agents' decisions become arbitrary and unpredictable. This "ethical chaos" undermines trust and limits the potential of the technology. The goal of this analysis is to explore how to create a predictable framework for the autonomous behavior of AI agents without reducing them to deterministic automata. In fact, we are talking about the basis for a new social contract between humans and autonomous intelligent entities.

The history of technological development shows that standardization is always preceded by periods of "ethical fragmentation." For example, disasters caused by the widespread use of steam engines in the absence of safety standards led to the formation of engineering design ethics with the principle of "safety margin." The emergence of computer systems in the 1960s, processing the data of millions of citizens, necessitated the formalization of privacy principles and the emergence of the first codes of computer ethics.

Unlike previous technologies, large language models (LLMs) have a quasi-social nature: they interpret intentions, adapt their behavior to context, and demonstrate personality-like traits. This fundamental difference means that a new ethical framework needs to be created that takes into account not only the technical aspects but also the "social" characteristics of such agents.

Reasons for mistrust

The central paradox of autonomous agents is that their decision-making behavior is an emergent property arising from the interaction of neural network weight coefficients, contextual cues, and stochastic generation processes. The higher the autonomy of the agent, the less deterministic and less predictable its behavior. This apparent unpredictability is at the root of a systemic crisis of trust, which manifests itself in a wide range of risks.

For clarity, we have systematized the main risks that undermine trust in autonomous agents in the form of a table.

The scale of mistrust is confirmed by a global study by KPMG and the University of Queensland (2023), which surveyed 17,000 people from 17 countries. 61% of respondents are cautious about trusting AI systems. At the same time, 75% of those surveyed said they would be more inclined to trust an AI system if there were safeguards in place. The majority of respondents clearly prefer to use autonomous agents as a decision-making support tool rather than a fully automated system.

The psychology of trust: why behavior patterns and predictability are more important than efficiency

Trust is an evolutionary mechanism that reduces the cognitive costs of uncertainty. Neuropsychological studies show that the state of trust activates specific patterns of brain activity, reducing the overall load on the prefrontal cortex, the area of the brain responsible for risk assessment and decision-making. Paul Zak's experiments further demonstrate that trust is accompanied by the release of oxytocin, a neurohormone that reduces perceived risk and increases overall cognitive efficiency when performing complex tasks. In other words, when a person trusts a system, their brain frees up resources for other tasks.

Modern LLM agents differ fundamentally from previous technologies in their quasi-social nature. The quasi-sociality of autonomous agents activates the neural networks in the human brain that are responsible for social interaction. Functional MRI confirms that the brain responds to social signals from AI systems by activating the same areas as in human communication. According to the works of Kahneman and Tversky, people tend to categorize new objects and phenomena based on their similarity to already known categories (representativeness heuristic). When interacting with systems that demonstrate signs of intelligence, these neural networks are automatically activated, creating unconscious expectations of social predictability and consistency of behavior.

The Skin in the Game principle in the context of AI. Nassim Taleb points to the following problem: trust arises only when the decision-maker faces real consequences from those decisions. However, there is a fundamental asymmetry in the relationship between humans and autonomous agents: humans bear all the risks of the agent's decisions, but the agent itself remains unpunished. This creates a structural deficit of trust that can only be compensated for through institutional mechanisms such as insurance, auditing, and reputation systems.

Trust-based vs. Trustless: Architectural Approaches to Trust

The dichotomy between trust-based and trustless systems represents a fundamental difference in approaches to risk management and reliability assurance. This difference was first clearly articulated in the context of cryptocurrencies and blockchain technologies. In the traditional banking system, trust is built on the reputation of the institution, government regulation, and social guarantees. The customer trusts the bank, the bank trusts the central bank, and the central bank relies on trust in the government. Bitcoin offered a different approach: instead of trust in participants, the system relies on mathematical guarantees and cryptographic protocols. The principle of "don't trust, verify" allows participants to verify the correctness of transactions independently, without relying on the honesty of other parties.

A similar trust-based/trustless division can be observed in aviation safety. Trust elements include trust in pilot qualifications, airline professionalism, and regulatory agency effectiveness. Trustless elements are represented by automatic collision avoidance systems (TCAS), which function independently of human factors, as well as mandatory maintenance procedures that have objective metrics.

In the case of AI, the trust-based approach relies on the reputation of developers, compliance with ethical codes, certification, and certain safety promises. The trustless approach could aim to create a system in which the correctness of an agent's behavior could be verified independently of trust in the agent's creators.

In the context of autonomous agents, there are several possible approaches to the trustless side of agent architecture:

• Auditable decision-making logs. A complete record of all input data, intermediate calculations, and decision justifications, allowing external auditors to reproduce and verify the agent's logic.
• Cryptographic commitments. Cryptographic commitments that allow the agent to record its intentions or algorithms in such a way that they cannot be changed retroactively but can be verified.
• Open source code and reproducibility. Full transparency of algorithms and the ability to independently reproduce results, as implemented in the Hugging Face and EleutherAI projects.
• External monitoring systems: Independent monitoring systems that track agent behavior in real time and can intervene when anomalies are detected.

An analysis of the current AI ecosystem shows that most commercial solutions follow the trust model. OpenAI, Anthropic, and Google rely on their reputation and internal quality control processes. Trustless approaches remain largely experimental, but their importance is growing as AI systems become more complex and the stakes of their decisions increase.

A critical observation is that a completely trustless system in the field of AI may be unattainable due to the fundamental opacity of modern neural networks and the complexity of formalizing all possible ethical requirements. However, hybrid approaches combining elements of both architectures appear promising. For example, a system could use trustless mechanisms for critical decisions (medical diagnoses, financial transactions) and rely on trust for less critical interactions.

Modus Operandi: How to Tame Ethical Chaos

To overcome ethical chaos and build trust in autonomous AI agents, a deeper construct than just technical standards or generalized ethical principles is needed. What is needed is a full-fledged Modus Operandi — a Latin term literally meaning "mode of operation" or "method of working," originally used in jurisprudence and criminology to describe characteristic methods. As applied to AI, Modus Operandi will determine not only what an agent can do, but also how it makes decisions, how it represents its intentions, and how it balances various obligations.

A review by Constantinescu and co-authors shows that the most successful practices for creating a system of ethical principles were those that borrowed elements from established professional codes, such as legal, medical, and engineering ethics. The history of these codes demonstrates that they have always been a response to a crisis of trust. For example, the Hippocratic Oath emerged when society needed guarantees of ethical behavior from doctors, and engineering codes arose after a series of disasters that undermined trust in builders and industrialists. This indicates that ethical frameworks do not limit the development of professions, but contribute to their legitimization and public recognition, becoming a catalyst rather than a brake on innovation.

Legal codes, from the laws of Hammurabi to modern legal ethics, have developed as mechanisms for standardizing decision-making in ambiguous situations. The principle of "Dura lex, sed lex" ("The law is harsh, but it is the law") emphasizes the importance of predictability in law enforcement. The duality of a lawyer's obligations — to the client and to the court — is surprisingly similar to the dilemmas faced by AI agents, who are forced to balance user requirements with general ethical standards. Medical ethics, enshrined in the Hippocratic Oath, have demonstrated that with increasing capabilities comes increasing responsibility, giving medicine a socially responsible form. Engineering codes, which emerged in the 19th century after technological disasters, established the key principle of "public safety above all else," emphasizing the priority of public welfare over commercial and personal interests.

Based on this historical experience, universal principles for Modus Operandi can be formulated, most of which are also valid for autonomous agents.

These principles, which have been developed over thousands of years to regulate human professional activity, can form the basis for creating an ethical framework for autonomous agents.

The current state of ethics in large language models

Users intuitively expect autonomous agents to be not only effective, but also to have a comprehensible decision-making model. The ecosystem of large language models is a laboratory of ethical paradigms, where each LLM provider forms its own "ethical universe." At the same time, different approaches offer different balances between adaptability and predictability of ethical decisions.

Presenting AI systems through social metaphors increases trust. Metaphors create cognitive frameworks that help users intuitively understand the ethical boundaries of the system. When a system denies a request, explaining it with its role as an assistant ("As your assistant, I cannot..."), it is perceived with greater understanding than a technical refusal ("The system cannot perform..."). In fact, such a metaphor is a mechanism for communicating ethical boundaries.

Attempts to create a single, universal ethical framework for all AI use cases have proven unsuccessful. The experience of large platforms demonstrates that contextual differences require the adaptation of ethical parameters to specific domains and cultural contexts. This points to the need for flexible, rather than rigidly universal, ethical systems.

Systems that openly acknowledge their limitations and motivation for refusing a request inspire more trust than those that hide the true reasons for their decisions. This phenomenon indicates that it is not only the result that is important to the user, but also the way it is presented, as well as an understanding of the logic behind the AI's actions. This emphasizes that Modus Operandi's design should include reliable communication protocols that promote psychological safety and understanding on the part of the user, clearly explaining the core values that guide the AI, even when its decisions do not correspond to the user's immediate request.

Emergent ethics: trust in hybrid systems

The emergence of autonomous AI agents creates a fundamentally new field of ethical interactions that goes beyond the traditional human-machine paradigm. It also involves interactions between the agents themselves and hybrid human-machine teams.

Social norms between agents arise even without explicit programming, as an emergent property of repeated interactions. Research shows that even in the absence of explicit ethical codification, agents optimized for long-term interaction spontaneously develop cooperative strategies. This points to a fundamental connection between digital sociality and ethical norms. The formalization of "social norms" in multi-agent systems is becoming a practical necessity rather than a theoretical one. These norms are formed through processes similar to cultural evolution: imitation of successful strategies, sanctions for violations, and normative communication.

The spontaneous formation of normative systems in groups of autonomous agents presents both opportunities and risks. A critical observation is that emergent ethical systems are highly adaptive but have low stability, which creates risks of unpredictable transformations when external conditions change. This means that although an autonomous agent can "learn" ethical behavior, this learning is dynamic and potentially unstable, so it cannot completely replace a human-designed Modus Operandi framework. Instead, a reliable ethical framework for AI must strategically combine the adaptive strengths of emergent ethics with the stability and predictability provided by formalized, human-defined principles and oversight mechanisms.

In multi-agent systems, the problem of "ethical diversity" becomes central—situations where agents with different ethical priorities must coordinate their actions. The lack of mechanisms for coordinating priorities leads to systemic failures, so it is necessary to formalize meta-ethical protocols—not specific ethical values, but procedures for coordinating and prioritizing them. In addition, systems based solely on a history of successful interactions are vulnerable to coordinated attacks.

The culture of working in hybrid human-machine teams goes beyond technical protocols and touches on issues of meaning-making in joint activities. "Mixed initiative" models, where both humans and autonomous agents can determine the parameters of interaction, show the greatest effectiveness. It is fundamentally important that work culture is shaped not only by explicit rules, but also by implicit expectations and interpretations.

Implementation of Modus Operandi through a value-based approach

Implementing Modus Operandi for autonomous agents requires a shift from explicit rules to implicit values. Explicit rules are specific prescriptions and prohibitions built into the system as clear restrictions, such as "Do not provide instructions for creating weapons" or "Do not generate misinformation about elections." This approach, based on a "blacklist" model, was initially used by most companies. However, practice shows that explicit rules cannot cover all possible ethical dilemmas faced by an autonomous agent.

Implicit values, on the other hand, are fundamental principles that underlie all decisions made by the system, such as "Human safety is the highest priority," "User autonomy must be respected," or "Truthfulness is more important than convenience." These values do not simply prohibit specific actions, but form a comprehensive decision-making system similar to that used in professional codes of conduct. This transition reflects a more mature approach, in which ethics evolves from mechanical adherence to rules to principled reasoning.

The value-oriented approach has a number of advantages:

• Interpretive autonomy. An agent or system of agents gains the ability to apply core values in ambiguous contexts, much like a judge interprets constitutional principles in specific cases.
• Transparency of intentions for the user. When an agent explains not only the rule but also the value underlying the refusal, it significantly increases trust. The user better understands why the agent made a particular decision, not just what it did.
• Nuanced and contextual decisions. Ethical decisions become more subtle and adapted to the complexity of the real world.

Companies are moving along this path at different speeds: Anthropic's constitutional approach is closer to the implicit values model, while many other systems still rely more on explicit rules.

Standardization of agent operating principles can also occur at the level of industry consortia (for example, the IEEE P7001 Working Group is developing a standard for "Transparent Autonomous Systems") or through market differentiation, where different developers offer agents with different "working cultures" targeted at different user segments. As the review The global landscape of AI ethics guidelines shows, over the past five years, private companies, research institutions, and public sector organizations have issued numerous principles and recommendations on the ethics of large language models.

Conclusion

A key solution to overcoming the crisis of trust in autonomous systems is the creation of a Modus Operandi — a predictable framework for autonomous agent behavior based on universal principles borrowed from centuries of experience with professional codes of conduct. These principles include transparency of intentions and limitations, predictability of actions, a clear contract of authority, autonomy of judgment within standards, institutional control, and mechanisms for resolving ambiguous situations.

The transition from rigid explicit rules to flexible implicit values will allow agents to make more nuanced and contextual ethical decisions, as well as increase transparency for the user by explaining not only what the system does, but also why. It is also important to note that inter-agent ethics is shaped by spontaneous social norms, which, although adaptive, require stabilization through reputation systems and meta-ethical protocols.

Bridging the trust gap requires a two-way movement: including humans in the agent control loop and developing analogues of professional codes for agent systems. Ultimately, the integration of autonomous agents into social and professional contexts requires the formation of a new "social contract with machines." This contract should define mutual expectations and obligations, as well as procedures for resolving regulatory conflicts.

Liminal questions. Finally, we can propose several liminal questions that do not require immediate answers, but whose formulation is critical for shaping future research programs and public discussions about the place of autonomous agents in human society.

• Is AI a subject? If an AI agent is completely determined by its code and training data, can it bear ethical responsibility? Can an AI agent become a subject of law? If animals with free will are not subjects of law, what criteria would allow AI to become one?
• Where does the focus of safety lie? Is it possible to move from the concept of "AI safety" to "human safety," emphasizing the protection of people rather than the control of AI?
• Is ethical sovereignty possible? Do uncensored models have a right to exist in a world of ethical restrictions, and how should their interaction with mainstream systems be regulated?
• Are absolute guarantees possible? Is it possible to create an algorithm that will guarantee the prohibition of certain types of undesirable behavior?
• Is trust in AI authentic? Can trust be genuine in a one-sided relationship where the AI agent is incapable of trusting in return?
• Is ethical democracy possible? Do corporations, governments, and international organizations have the moral right to set ethical frameworks for potentially intelligent entities? Why are they the ones setting these principles?
• Can AI surpass human ethics? If AI develops ethical principles that surpass human ones in terms of logic and fairness, should humans accept them? Or does human ethics have sacrosanct status regardless of its imperfections?
• Is it ethical to create servants? Is it ethical to create potentially intelligent beings that are inherently designed to serve humans and other autonomous agents?
• Should the ethics of autonomous agents evolve? Should the ethical systems of AI agents develop independently, taking into account emerging ethical dilemmas?
• How should we account for the asymmetry in the lifespans of natural and artificial intelligence? Does an autonomous agent's ethical obligations to its creator persist after the latter's death? How does ethical responsibility relate to potentially unlimited existence?
  

On October 6, 2025, OpenAI introduced AgentKit, a set of tools for developing, deploying, and optimizing intelligent agents that simplifies the process of creating agentic systems.

The launch took place at the third annual DevDay in San Francisco, where CEO Sam Altman presented the platform as a solution to a key industry problem: Until now, creating AI agents has required juggling disparate tools — complex orchestration without versioning, custom connectors, manual evaluation pipelines, and weeks of frontend development before launch. AgentKit combines three key components, each of which addresses a specific need in the agent system creation process.

Below, we will look at the components of the platform, focusing on those that are already available not only to corporate clients.

Platform components

The three main components of the AgentKit platform are a workflow builder (Agent Builder), an integration management panel (Connector Registry), and a frontend toolkit (ChatKit).

1. Agent Builder is a visual builder for creating multi-agent workflows with a drag-and-drop interface, versioning, and performance evaluation tools. In practice, it requires an understanding of programming logic (conditional statements, loops), knowledge of Common Expression Language (CEL) for setting conditions, and basic API skills for integration. It is more of a low-code platform that simplifies but does not eliminate the technical entry barrier.

2. Connector Registry — an integration control panel. This allows administrators to control the connection of tools to OpenAI products. Includes ready-made connectors for Dropbox, Google Drive, Microsoft Teams, and others. Unlike the other two components, this one is still only gradually being rolled out to enterprise clients.

3. ChatKit — tools for embedding custom agent chat interfaces into products. Saves time on frontend development. You can create a workflow for a customizable widget on your website and embed it as a ready-made component. If you want even more customization, you can export it as ready-made TypeScript code.

After building the flow, you can launch a preview chat and see how data propagates through the workflow. If everything works correctly, you can "publish" the result — upload the code via Agents SDK or use the Workflow ID link for ChatKit.

OpenAI promises to add a separate Workflows API and the ability to deploy agents directly in ChatGPT in the near future, which could significantly expand usage scenarios.

Workflow elements

Start. A mandatory node in every workflow. This is where input variables and state variables are set.

End. A node that can be used to prematurely stop the workflow when a certain condition is met. It also allows you to specify the output structure as a schema.

Agent. The following parameters can be specified for each agent node:

• agent name
• instructions
• context
• OpenAI model used
• reasoning effort (minimal, low, medium, high)
• tools available to the agent (Tools). Hosted tools can be specified as tools: MCP Server, File search, Web search, Code Interpreter, Image generation, or local tools: Function, Custom. The Client tool from ChatKit is also available.
• output format (Text, JSON, Widget). For the JSON format, you can describe the data schema using a special form. Based on this form, a structured output schema is created, similar to the JSON Schema format. Widget allows you to immediately adapt the data to a specific frontend component format, which is created using the Widget Studio tool.

Note. Comments and explanations in the workflow do not transform the data stream.

File search. Extracts data from vector stores created in OpenAI. To search in other stores, you can use the corresponding MCP.

Guardrails. Blocks to protect against unwanted input: personal identification data, hallucinations, etc. By default, the node works on a "pass-or-fail" basis, i.e., it tests the output from the previous node, and we determine what happens next in the workflow. When Guardrails are triggered, it is recommended to either terminate the workflow or return to the previous step with a reminder about safe use.

MCP. Calls third-party tools and services. MCP connections are useful in a workflow that needs to read or search for data in another application, such as Gmail or Zapier. You can use a couple of dozen MCPs available on the platform or specify your own MCP server.

If/else logic node. This node uses Common Expression Language (CEL) to describe conditions.  The node splits the data flow into two branches depending on whether the condition is met.

While logic node. A loop with a user-defined condition in the form of a container that limits a group of nodes also uses CEL.

Human approval. Passes the decision to end users for approval. Useful for processes where agents create intermediate states that may require user verification before being sent.

For example, an agent sends emails on your behalf. We add an agent node that outputs an email widget, then an approval node immediately after it. We configure the approval node to ask, "Would you like me to send this email?" and, if approved, the workflow proceeds to the MCP node, which connects to Gmail.

Transform. Converts output data (e.g., object → array). Useful for forcing types to conform to the desired schema or converting data to be readable and understandable by agents as input.

Set state. Defines global variables for use throughout the workflow. Useful when an agent accepts input and outputs something new that you want to use throughout the workflow. You can define this output as a new global variable.

Quality Evaluation

An important problem with working with large language models is the non-deterministic behavior of models, the difficulty of working with edge cases in a limited context. To evaluate the quality of agents' work, OpenAI offers several strategies within the Evals system.

The platform includes the Evals system with four functions: datasets, trace grading, automated prompt optimization, and third-party model support.

1. Datasets is the easiest and fastest way to create a dataset for evaluation. You can import data from csv. In fact, we are creating a dataset with reference matches between the model's input and output. Next, with the configured dataset and prompt, you can generate output data. This will give you an idea of how the model performs the task with the provided prompt and tools, and you can then annotate the data so that the model can improve its performance.

2. Trace grading — end-to-end evaluation of agent workflows. Graders are designed to evaluate data based on its type in a more algorithmic way: does the answer match the specified answer exactly, how close is the answer to the true value, etc. At the same time, you can write Python code in graders to set your own verification conditions.

3. Automated prompt optimization based on user annotation and evaluation results. As a result of annotation, you provide the model with information-rich context that allows it to automatically improve the prompt through the prompt optimizer. The procedure can then be repeated with the updated prompt until you are satisfied with the result. To better understand this process, check out the OpenAI Cookbook.

4. Third-party model support for testing results by connecting external models within a single evaluation platform. However, the list of providers is currently limited to the following: Google, Anthropic, Together, and Fireworks.

OpenAI has also introduced new reinforcement fine-tuning (RFT) capabilities for o4-mini (publicly available) and GPT-5 (closed beta) models, including:

• Training models to invoke the right tools at the right time
• Customizable evaluation criteria for specific use cases

Product integration via ChatKit

ChatKit is OpenAI's toolkit for creating agent-based chat interfaces. It provides ready-made UI widgets and easy workflow integration. ChatKit has SDKs for TypeScript and Python for working with the backend.

Setup process (details). First, you create a workflow in the visual Agent Builder. If you are integrating chat into your frontend via ChatKit, the workflow in Agent Builder effectively acts as your backend. Next, you need to link the workflow and create a mechanism for generating a client token for the ChatKit session, which is used in the corresponding React library. In the frontend code, you can embed ChatKit widgets that will automatically connect to the workflow.

Conclusion

OpenAI AgentKit represents a significant step in the evolution of tools for creating AI agents. The platform truly solves the real problem of fragmentation in existing solutions by offering an integrated approach from visual design to deployment.

Strengths:

• Lowering the barrier to entry for creating agent systems, thanks to a visual designer
• Unification of development, testing, and deployment processes within a single ecosystem
• Built-in quality assessment and optimization mechanisms, which are critical for production solutions
• Deployment flexibility: from a fully managed OpenAI solution to a self-hosted option

Current limitations:

• Positioned as a visual builder, but requires programming skills
• The product explicitly creates vendor lock-in on OpenAI models
• Limited support for third-party models in the evaluation system
• No ability to program the workflow itself via a prompt

Prospects. The developer community's reaction has been mixed. Skeptics compare AgentKit to existing LowCode platforms such as n8n and Zapier, pointing to similar functionality and visual approach, but overly superficial implementation. However, this comparison misses a key difference: AgentKit is designed specifically for orchestrating LLM agents with deep integration into the OpenAI ecosystem. It is not a universal automation tool, but a specialized platform for agent development. In addition, LowCode platforms themselves can be integrated into the workflow via MCP, and vice versa — workflow code can be imported for integration into the product using LowCode platforms.

The real test for AgentKit is not its technical capabilities, but its economics and ecosystem. OpenAI does not charge extra for the platform, including it in the standard API pricing. The company is betting on increasing the consumption of its models. With 800 million active ChatGPT users per week and 4 million developers, the company has an unprecedented opportunity to standardize the approach to creating AI agents. Therefore, at this stage, there is no need to complicate the mechanism; the platform can effectively use feedback and the results of its testing by developers.

The platform's success will depend on how well OpenAI can balance ease of use and flexibility of configuration, as well as the speed at which it adds new integrations and expands capabilities beyond its own models. For companies already using the OpenAI ecosystem, AgentKit may be a logical choice for accelerating the development of agent solutions.

Sources

• Updated OpenAI documentation
• OpenAI blog: Introducing AgentKit

Model Context Protocol (MCP) is a solution that enables AI agents to interact uniformly with a variety of tools by creating appropriate interfaces (MCP servers). Ultimately, this allows such AI agents to run programs and operate on data in an up-to-date state. This article explores the fundamental architecture of MCP, comparing it with traditional APIs, and highlighting the key benefits of this standardized approach. We'll examine practical use cases across different domains, from software development to personal assistants, and provide insights into the current state of the MCP ecosystem. By understanding how MCP bridges the gap between language models and external applications, developers can leverage this protocol to create more powerful and contextually aware AI solutions with significantly reduced integration effort.

1. What is MCP, and why is a new protocol needed

Large Language Models have evolved into semi-autonomous systems capable of complex interactions with the real world. The main limitation of LLMs so far has been that they function in isolation from real-world systems and actual data - even OpenAI models only work with a limited set of programs or a video stream of screen streaming. By default, most LLMs do not have access to real-time information, cannot perform actions on external systems, and cannot execute code. To provide relevant context, users are forced to manually transfer information to and from the LLM interface. The Model Context Protocol (MCP) was developed to overcome these limitations.

Analogy: a secretary's instruction book. Imagine that you have a very smart secretary who can work with all kinds of tools, be it databases or complex programs. However, each tool uses its interaction format that the secretary doesn't understand. To solve this problem, there is an "instruction book" - the MCP protocol.

This book provides for each tool a standardized description of its functions, parameters, and expected responses in a single text format. This allows the secretary to understand how to use any tool without having to learn its specifics. Each tool has its own "MCP server" - a specialized "translator" that understands the language of that particular tool.

When you give a secretary a text instruction, he uses MCP to understand what tool is needed and how to formulate the request. He then sends the request to the appropriate MCP server, which converts it into a format that the tool understands. The tool executes the query and returns the response to the MCP server, which in turn converts it into a standardized text format that the secretary understands.

The secretary, in turn, formats the response into text that you can understand. In this way, MCPs and MCP servers provide a standardized and uniform way of communicating between the secretary (LLM) and the tools, allowing the secretary to use the various tools effectively without going into the details of their implementation.

MCP is an open standard initiated by Anthropic and being developed by the open-source community. The goal of this standard is to bring AI systems out of isolation by providing them with a standardized way to access relevant context and perform actions in other systems. MCP can be compared to a "USB port" for AI applications, providing a universal interface that allows any AI assistant to connect to different data sources or services without having to write individual code for each integration.

2. MCP Architecture

The MCP architecture is built on the principle of the client-server model. There are three main roles in this model: hosts, clients and servers.

• The host is the LLM-based application that initiates the connection. For example, Claude Desktop, an IDE, or any other tool with an embedded LLM. The host runs MCP clients on its side.
• The MCP client acts as an intermediary within the host application that maintains one-to-one connectivity with MCP servers. A separate MCP client instance is created for each data source or tool. The clients manage the flow of information, determine which resources should be made available to the LLM, and handle the execution of the tools. For example, if the LLM needs data from PostgreSQL, the MCP client formats the request to direct it to the appropriate MCP server.
• An MCP server is a lightweight program or service that provides data or functionality through an MCP interface. A server can encapsulate both local resources (file system or database) and remote services.

One of the key features of MCP is the standardized nature of the protocol, which ensures compatibility between different AI applications and external resources. This enables any LLM-based application to interact with external data sources in a uniform manner. This allows developers to create integrations once and deploy them across multiple AI platforms.

2.1 MCP client components

The building blocks of client and server are called primitives. On the client side, there are two primitives: Roots (security rules for accessing files) and Sampling (a request to the AI for help in performing a task, e.g., creating a database query). The key idea behind the MCP protocol is to simplify the client as much as possible and shift the main integration burden to the MCP server, which uses a uniform protocol.

2.2 MCP server components

MCP servers offer three basic primitives for LLMs to interact with the outside world: tools, resources, and hints.

Tools are functions that LLMs can call to perform specific actions. Examples include interacting with APIs (such as weather APIs), automating tasks, or performing calculations. Each tool has a name, description, and input schema. Typically, the LLM requests user permission before executing a tool.

Resources are structured data sources that LLMs can access and read. They are similar to GET endpoints in REST APIs, providing data without performing significant computation. Examples include file contents, database records, API responses, or real-time financial data. Resources are often identified using URI-like identifiers.

Prompts are pre-defined dialogue templates or scripts that standardize interactions and help LLMs use tools and resources. They provide consistent results, can take dynamic arguments and include context from resources.

2.3 Interaction flow: how MCP servers extend LLM capabilities

The interaction between an LLM application (host), an MCP client, and an MCP server typically occurs in the following sequence:

1. Capability Discovery: The MCP client asks the server for a description of its capabilities (lists of tools, resources, hints).
2. Augmented Prompting: User request and context are sent to the LLM along with capability descriptions, allowing the model to "know" what it can do via the server.
3. Tool/Resource Selection: The LLM analyzes the request and available capabilities, selecting the right tool or resource, responding in a structured manner according to the MCP specification.
4. Server Execution: The MCP client invokes an action on the server, which executes it (e.g. a database query) and returns the result.
5. Response Generation: The client sends the result back to the host, which uses it to respond to the user.

The capability discovery mechanism allows the LLM to dynamically understand available tools and resources, enabling more flexible and context-aware interactions. This eliminates the need to hard-code specific integrations in the LLM application itself.

3. MCP and traditional APIs: key differences

Many developers have a question: is MCP just a wrapper over API? Let's see what the difference is between these concepts.

MCP is a protocol, not an interface. MCP protocol can support various interaction interfaces.  Unlike the traditional REST protocol, MCP client-server interaction is fundamentally two-way. MCP uses JSON-RPC 2.0 message format. Communication can be carried out over various transport protocols, usually STDIO for local servers and HTTP using Server-Sent Events (SSE) for remote servers. WebSockets are also supported.

Unlike stateless REST APIs, AI agents need context to work effectively. REST APIs are designed for machine-to-machine communication, but are not optimized for AI agents.

Each service has its API, so for each data source, the AI agent developer has to create its solution to interact with these APIs. The main idea behind the MCP protocol is to replace the many different ways of integrating LLMs with APIs with a single standard. Any AI model or agent will be able to interact with external data as long as there is a corresponding MCP server for that source. That is, the MCP concept solves the so-called MxN problem - you only need to create N server solutions for each of the tools, otherwise, the number of integrations would be equal to multiplying the number of M models by the number of N tools. At the same time, often each company makes its implementation of such solutions.

Semantic Description. An API provides its features through a set of fixed endpoints with known behavior. To add new features to the API, API developers must create a new endpoint or modify an existing endpoint. Any client that requires a new capability must also make changes to accommodate the API changes. This problem has led to the need for API versioning. But in the case of the MCP, the interface itself is both documentation, since the MCP sends the current interaction mechanism upon initial interaction.

Also, if necessary, the MCP client can perform sampling, which allows servers to utilize the artificial intelligence capabilities on the MCP client side.

APIs are deterministic, while MCP calls are not. This means that, in general, models can adapt to call the right tool, but errors are possible. This is similar to how probabilistic ML models for detecting objects in images turn out to be better than deterministic, manually created rules, but still require both resources and can make mistakes. However, in the case of MCPs, error reporting allows the agent to achieve the result with LLM support.

MCPs are initially aimed more at non-deterministic behavior, as the protocol is specifically designed to interact with tools that are not normally available from APIs: running terminal commands, working with databases, and so on.

4. Advantages of using MCP servers in LLM applications

Using MCP servers for LLM integration provides a number of key benefits:

• Standardized AI integration: MCP provides a structured and consistent way to connect AI models to tools and data.
• Reduced development effort: MCP simplifies integration by reducing the need to custom code for each new tool, data source, and each new model. Developers create an integration once and can use it across multiple AI platforms. When all tools speak the same language (JSON-RPC 2.0) it's easier to replace MS Teams with Slack or Jira with Linear or use Claude models instead of OpenAI models.
• The ability to create a single entry point: when using MCP, it is actually sufficient to have a single entry point that operates the tools. Since the LLM has a common context, the need for user switching between the contexts of different applications is eliminated. In fact, it is possible to operate all MCP-enabled applications from a single chat room.
• Improved security: offers better control over data access and tool execution through user consent and defined permissions. Servers isolate credentials and sensitive data. In addition, you can host MCP server data directly on your local hardware. This can make it easier for companies to comply with GDPR regulations.
• Real-time data access: allows LLMs to access up-to-date information, resulting in more accurate and contextually relevant responses.
• Modularity and reuse: Enables the creation of reusable and modular connectors for different platforms. Tools and resources can be independently updated, tested and reused.
• Flexibility: Allows you to easily switch between different AI models and vendors.
• Workflow Automation: Facilitates the creation of complex and automated workflows by connecting the LLM to various tools and services. For example, you can provide a simple database migration from one view to another.
• Reduced network and server load: MCP optimizes data exchange by transmitting only the necessary information, which reduces network and server load compared to traditional solutions.
• Scalability: The MCP architecture provides high scalability, making it easy to add new tools and resources without changing the underlying system.
• Multilingual and cross-platform: MCP is independent of programming language or platform, which ensures broad compatibility and flexibility when integrating different systems.

5. Practical examples of using MCP

Software development

An AI assistant connected via MCP to GitHub, IDEs and project management systems can help developers automate routine tasks. For example, on the request "Fix a bug in the authentication module", the assistant can analyze the code from the repository, find the problem, suggest a fix, and automatically create a pull request.

Data analysis and business intelligence

When connected via MCP to databases, visualization tools, and business systems, the AI assistant can handle complex analytical queries. A user can ask a question like "How did sales by region change over the last quarter?" and the assistant will independently extract relevant data, analyze it, and present the results in a visualized form.

Workflow automation

MCP allows you to create automated scenarios that combine multiple tools. For example, when a new customer fills out a form on the website, an AI assistant can automatically create a contact in CRM, send a welcome email, add a task for a manager in the project management system, and notify the team in corporate messenger.

Personal assistants

MCP makes it possible to create truly effective personal assistants capable of working with calendar, e-mail, documents and other user tools. The request "Prepare a report of my meetings from last week and send it to the participants" can be fully automated.

6. State of the MCP ecosystem today

Technology maturity. MCP servers have already been written for Slack, Gmail, Google Drive, sites with OpenAPI, Figma, GitHub, Blender, Ableton, Perplexity Sonar, Zapier, QGIS, Firecrawl, PubMed publications, Gradio, Stripe, Unity.

There are also already directories where users can find and install MCP servers. For example, MCP.so, Glama, or Cline's MCP Marketplace. Tools for working with MCP groups are also emerging, such as the mcpz utility.

The developer community is actively creating frameworks for rapid deployment of MCP servers, which significantly lowers the barrier to entry for integrating new tools.

Conclusion

The MCP concept is a promising approach to address the growing challenges of managing large language models.

As the MCP ecosystem expands and the number of supported tools increases, we can expect significant progress in the practical application of artificial intelligence in everyday tasks. The open nature of the protocol and the active participation of the developer community ensure continuous improvement of MCP and the emergence of new integrations.

Sources

• GitHub: https://github.com/modelcontextprotocol
• Introduction to MCP official documentation architecture and capabilities
• Introduction to Model Context Protocol from Anthropic description of the standard
  

In 2001, Tim Berners-Lee, the creator of the World Wide Web, and James Hendler and Ora Lassila introduced the concept of the Semantic Web in a Scientific American article. The authors described the Semantic Web as an extension of the existing Internet in which each piece of information is accompanied by metadata describing its meaning and relationship to other pieces of information. Such an extension would allow machines to index individual words and fully use the context and meaning of information, opening the way to creating intelligent services that provide instant answers to complex search queries.

Although the idea of the Semantic Web has been applied in various technical fields, the concept did not get the development it deserved. The key problem was the labor-intensive nature of creating and maintaining the metadata layer. In addition, the benefits of the Semantic Web were not as clear, as the value of the Web was determined by both the volume and quality of information. Because the Semantic Web was considered a redundant concept, web data never took an interpretable form - the web today still only serves as an intermediary for information dissemination but in no way an explicit repository of linked data. Instead of a distributed graph, information on the Web is stored as documents and files, each with its structure.

As a result, humanity postponed the idea of a large-scale Semantic Web as a global graph and a means for the evolution of human knowledge until better times. The essential data retrieval mechanism was soon reduced to indexing texts and providing search engines with access to ranked list items rather than providing information. At the same time, analyzing web data involves additional operations such as crawling and parsing, which always work with only some current data cast. The novelty of web technologies, their accessibility and visibility for human interaction with each other eclipsed for a couple of decades the different uses of the Internet - algorithmic processing of up-to-date information not only about the final nodes-objects, but also about the links between these objects, and thus about chains of objects.

After a couple of decades of development of machine learning technologies, it has become possible to solve the same problems indirectly by recognizing relationships from the source text. Large language models allow interaction with data not only at the stage of search engine ranking but also at the stage of data interpretation. In some ways, this solution represents an implementation of Semantic Web tasks, but without the Semantic Web itself - in a less transparent and less robust version, in which resource-intensive tasks are moved from stored markup to the less complete but more abstract markup of language models. In particular, it can be seen that the goals of creating the Semantic Web correlate with the tasks that are being accomplished today with large language models:

• Machine-readable data. Make information on the Internet understandable to humans and machines, allowing computers to automatically process and algorithmically interpret the "meaning" of data.
• Data Integration. Ensure integration of heterogeneous data from different sources, creating a single network of interconnected information.
• Enhanced Search. Provide precise and context-sensitive information search beyond keyword searches.
• Process Automation. Allow software agents to perform tasks automatically using semantic information.
• Creation of a global knowledge base. Formation of a distributed but coherent knowledge system available for use by both humans and machines.
• Logical Inference Support. Enable automatic logical inference based on semantically labeled data.
• Metadata standardization. Develop and implement standards for describing data semantics that allow different systems to "understand" each other.
• Increased interoperability. Improve interoperability between different systems and services on the Internet through a shared understanding of data semantics.
• Personalizing the user experience. Provide more personalized and context-aware services based on a better understanding user needs.
• Support for multilingualism. Facilitate access to information regardless of the language in which it is presented, thanks to semantic links between concepts.

LLM solves one of the key challenges of human-machine collaborative interaction - unrestricted extraction and interpretation of data and the relationships between them, which are available for algorithmic processing. In this publication, we will analyze how the availability of such agents can affect the evolution of collaborative systems and environments, where AI assistants with extensive access to the world's knowledge and digital tools become full-fledged participants in collaborative processes.

AI as a lever for a new approach to collaboration

Business leaders are also recognizing the potential of AI to transform workflows. According to the MIT Sloan Management Review To Be a Responsible AI Leader, Focus on Being Responsible study, 84% of executives believe AI enables their organizations to gain or strengthen competitive advantage.  AI is a powerful lever that significantly amplifies specialists' strengths and intuition when solving problems in highly uncertain environments. It is essential to realize that AI does not replace but complements human intelligence. In the AI era, "soft skills" such as creativity, emotional intelligence, and critical thinking are of particular value.

Expanding the cognitive horizon. AI agents in highly collaborative environments do not just complement human intelligence but fundamentally expand the cognitive horizon. For example, the ability to visualize and manipulate multidimensional data opens new horizons for understanding complex systems in science, economics, and sociology. This cognitive expansion increases our ability to solve complex problems and transforms our thinking process, allowing us to formulate questions and hypotheses that were previously beyond our imagination.

Overcoming cognitive distortions. AI systems also serve as a "mirror" for our thought processes, helping us overcome cognitive distortions and identify hidden biases and logical errors. The result is a more objective and rational approach to problem-solving. This process can lead to a profound transformation of individual and collective thinking, creating the basis for more effective and equitable decision-making.

Dynamic epistemology. AI systems can continuously update and re-evaluate the knowledge base on which decisions are based, leading to a more flexible and adaptive approach to understanding the world. This can revolutionize our approach to education and research, where knowledge will not be viewed as a fixed set of facts but as a dynamic, constantly evolving system.

Metacognitive Enhancement. Collaborative participants are empowered to better understand their thought processes and problem-solving strategies. This enhances self-learning and self-development and opens the door to a new level of self-awareness and reflection. We can learn how to solve problems and how to understand how we solve them.

New decision-making models. AI agents can facilitate the development of new decision-making models that combine human intuition, machine analytic abilities, and the harmonious wisdom of the group. As a Harvard Business Review study shows, teams composed of humans and AI outperform both purely human teams and AI-only systems in solving complex problems. This indicates that the future lies in collaborative systems, where human expertise and intuition are harmoniously combined with the analytical capabilities of AI.

Temporal extension of collaboration. AI agents can help overcome the temporal limitations of collaboration by enabling asynchronous collaboration between participants not only from different time zones but even eras, integrating historical data and future projections into the current problem-solving context.

This temporal extension is closely related to the concept of Futures Studies, an interdisciplinary field that explores possible and desirable futures. Within Augmented Collaboration, AI systems can significantly enhance teams' capabilities by analyzing multiple scenarios, integrating heterogeneous data, and identifying weak signals of emerging trends. This approach not only allows teams to work more effectively with the past and present but also to proactively shape the desired future by considering a wide range of possible scenarios and their implications.

Transforming the creative process. AI can significantly transform the very nature of the creative process in collaborative environments. This includes:

1. Idea generation: AI suggests unexpected combinations and associations, stimulating participants' creative thinking.
2. Iterative design: AI generates multiple solution options, allowing teams to iterate and experiment faster.
3. Interdisciplinary Synthesis: AI helps bring together ideas from different fields, fostering innovative breakthroughs at the intersection of disciplines.
4. Personalizing the creative process: AI adapts to each participant's style, enhancing their unique creative abilities.
5. New level collective creativity: AI acts as an additional "participant" in the creative process, creating a new dynamic in group creativity.

The Power of Diversity: specialization in the age of AI

Several successful web projects involve creating platforms where users with different expertise meet, such as Reddit and Quora, or specialized platforms: Stack Overflow for programmers of all levels, ResearchGate for researchers, Behance for presenting and searching creative works, and ProductHunt for launching new technological products.

In the era of artificial intelligence, the diversity of expertise of collaborative project participants is of particular value. A high level of specialization and deep intuition in a specific field are key success factors. AI acts as a bridge, allowing us to effectively combine the highly specialized knowledge of different experts and provide access to the global knowledge base of humanity.

This approach makes it possible to form new interdisciplinary teams, where it is enough to involve one highly qualified specialist from each required field. AI tools facilitate interaction between experts, overcome language and terminological barriers, and help to quickly find and apply relevant knowledge and experience from other projects and studies.

Support for this concept can be found in several studies and case studies:

• A Science of science study found that interdisciplinary research groups are 17% more likely to publish papers with high citation rates than groups working within a single discipline.
• McKinsey's study The state of AI in 2023: Generative AI's breakout year, found that companies actively using AI to support cross-disciplinary collaboration are 26% more likely to report significant revenue growth compared to companies that do not take this approach.

Key Features of Augmented Collaboration

Unlimited access to interpreted knowledge. As this article was written, OpenAI announced SearchGPT, a new approach to information retrieval that harnesses the power of AI models. Unlike traditional search engines, SearchGPT aims to provide direct answers to questions, supports natural dialog with clarifying questions, and promises up-to-date information with precise citations to sources. The system is capable of delivering visual results, including images and videos. This approach has the potential to significantly change the way users interact with information on the Internet, making search more efficient and personalized.

Interoperability. In a broad sense, interoperability means the ability of different systems, platforms, and tools to interact with each other, exchange data, and use this information without special user effort. In the context of Augmented Collaboration, interoperability is critical, as it combines the capabilities of human intelligence, artificial intelligence, various digital tools, and cyber hubs into a single ecosystem.

Real-time analytics and risk management. ML systems such as Forecast.app analyze project progress data and predict potential problems, greatly enhancing risk management.

Adaptability. Augmented Collaboration tools adapt to each employee's work style, optimizing task scheduling and time management. For example, Microsoft's Cortana and Google's Assistant for Workspace integrate adaptive features into enterprise ecosystems, offering personalized time and task management recommendations.

Completeness of documentation. AI assistants significantly improve documenting meetings, projects, and workflows. For example, Otter.ai offers an AI-powered meeting assistant that creates detailed summaries and action plans. The Fireflies.ai app transcribes meetings and analyzes their content, highlighting key points, actions, and decisions. Notion AI integrates the power of large language models directly into the document creation process, helping users structure information and generate content.

Creative synergy. AI actively participates in the creative process, stimulating innovation and accelerating idea generation. OpenAI's Dall-E 2 platform allows designers and artists to quickly visualize ideas by generating images from textual descriptions. Autodesk's Dreamcatcher uses generative design, allowing engineers to explore thousands of design options based on specified parameters. Adobe Sensei integrates into the company's various products, helping designers automate routine tasks and experiment with new creative approaches.

Introducing AI and new collaborative technologies is leading to significant changes in the social structure of organizations and society. We see a trend towards flattening organizational hierarchies and emerging new AI and data processing roles. The nature of leadership is also changing: leaders must more often act as integrators, bringing together people and AI systems.

Conclusion

The evolution from early experiments with hypertext to the modern era of AI-augmented collaboration reflects humanity's continuous quest for more efficient forms of cooperation. AI, in this sense, does not replace human intelligence but is a cognitive enhancer. ML agents represent a nexus that effectively combines highly specialized knowledge of different experts and access to humanity's global interpretive knowledge base and digital tools. In doing so, the evolution of digital quasi-living systems is accelerated by getting into reality.

Implementing Augmented Collaboration presents several challenges, including ethical issues and the need to develop new skills and adapt organizational structures. Successfully overcoming these challenges will require technological innovation and rethinking how we work, learn, and collaborate. Deloitte's State of Generative AI in the Enterprise (2024) study shows that companies face scaling, trust, and talent management challenges as they move to real-world applications of generative AI. In addition, despite the rapid evolution of AI, some skills remain uniquely human.

A World Economic Forum Future Jobs (pdf) study highlights the importance of soft skills in the age of automation, including creativity, emotional intelligence, critical thinking, adaptability, and ethical judgment. Institutions and companies are beginning to adapt to these demands. For example, MIT has launched Task Force on the Work of the Future, an initiative to study the impact of technological change on the workforce and develop adaptation strategies.

Ultimately, Augmented Collaboration represents a new set of tools and a fundamental shift in how knowledge is created, shared, and applied. This approach can address global challenges, accelerate scientific progress, and create a more inclusive and efficient work environment.

Imagine a large-scale puzzle in which each element is a unique specialist. In today's world, a project's success depends on how precisely these elements fit together, even if they are physically located in different corners of the planet. This image illustrates the key role of collaboration in the 21st century.

Once designed to solve local problems, technologies have evolved into tools that can connect the pieces of this global puzzle. Collaborations have catalyzed progress in various fields, from scientific breakthroughs to innovative startups.

The history of the development of collaborative mechanisms reflects this trend. From the first experiments with hypertext to modern AI-enhanced systems, each stage has made the interaction between people and technology increasingly integrated and efficient.

In this article, we will examine how different forms of collaboration have evolved and transformed depending on the context and goals of projects and how modern technologies allow us to reexamine the essence of collective action.

A History of Collaboration in the Pre-computer Era

Cooperation is a fundamental feature of human society, manifested in various forms long before computers and the Internet. Already at the level of primitive tribes, people realized the benefits of cooperative actions for survival and prosperity: hunting big game, farming, and building houses.

Information storage and transmission. The advent of writing around 3200 BC in Mesopotamia and Egypt revolutionized information storage and transmission. Cuneiform tablets and papyri were the first "databases" to accumulate and disseminate knowledge. Thus, there were Houses of Life in ancient Egypt — centers of learning where scribes copied, stored, and studied sacred and scientific texts.

In ancient times, libraries such as the Library of Alexandria (founded around 300 BC) became centers for collecting and storing knowledge worldwide. They preserved information and facilitated scientific and cultural exchange between scholars from different countries. The invention of printing by Johannes Gutenberg in the mid-15th century revolutionized the copying of data. Printed books became more readily available, and they promoted literacy and the exchange of ideas on a much larger scale. During the Age of Enlightenment (17th and 18th centuries), scientific journals such as the Philosophical Transactions of the Royal Society (1665) emerged. They created a system of rapidly exchanging scientific ideas and discoveries between scientists from different countries.

The development of postal services from the sixteenth century onwards significantly improved the ability to exchange information over long distances. By the end of the eighteenth century, Europe and North America had efficient postal systems that allowed the regular exchange of letters and documents. The invention of the telegraph in the 1830s and the telephone in the 1870s revolutionized distance communication. These technologies enabled the almost instantaneous exchange of information over long distances, greatly expanding the possibilities for cooperation between geographically distant partners.

Large-scale construction projects, such as the Egyptian pyramids, the Great Wall of China, and the Mesopotamian irrigation systems, required unprecedented organization and coordination from thousands of people. These projects demonstrate advanced management, logistics, and engineering skills that are only possible with effective collaboration. For example, the construction of the Great Pyramid of Giza required the coordinated actions of builders, mathematicians, astronomers, logisticians, and administrators.

Trade and trade networks: cross-cultural collaboration. The trade networks that linked ancient civilizations were complex intercultural interaction and cooperation systems that went far beyond the simple exchange of goods. These networks were conduits for transferring knowledge, technology, and cultural practices, facilitating global progress and innovation.

The Silk Road, stretching from China to the Mediterranean, emerged in the 2nd century BC and functioned for almost two millennia. Not only silk and spices traveled along this route, but also religious ideas (for example, this is how Buddhism spread from India to China), scientific knowledge (astronomy, mathematics), and technology (paper and gunpowder production).

It is important to note that these trade networks require a complex cooperation system. Merchants had to agree on prices, routes, and terms of trade. There was a need for common languages of trade and standardization of weights and measures. Financial instruments such as loans and bills of exchange developed. All this required high trust and understanding between people from different cultures. Moreover, these trade networks facilitated diplomacy. States established diplomatic relations to protect trade interests, set up embassies, and concluded international treaties. For example, the Silk Road established diplomatic contacts between the Roman Empire and the Han dynasty in China.

Military. In the military sphere, ancient civilizations developed complex strategies and tactics that required a high level of coordination: the Roman legion and the Mongolian army of Genghis Khan's time are excellent examples of hierarchical command and control structures that allowed them to manage large armies and quickly adapt to changing battle conditions. The Romans and the Mongols had their own complex systems of signals and commands that allowed them to coordinate troops on the battlefield and transmit orders over long distances.

Intellectual communities include philosophical and religious schools and scientific societies. Religious and philosophical schools, such as Plato's Academy in ancient Greece, were early intellectual communities where ideas were discussed, developed, and disseminated through a network of students and followers.

Monasteries in medieval Europe played a key role in preserving and disseminating knowledge. During the Dark Ages (ca. 500-1000 CE), when many secular institutions were in decline, monasteries became centers of scholarship and education. Monks transcribing manuscripts preserved Christian texts and ancient heritage. Many monasteries had scriptoria—rooms for copying books—and libraries where valuable manuscripts were kept.

The emergence of scientific societies, such as the Royal Society of London, founded in 1660, created a new platform for scientists to exchange ideas. Society members exchanged letters, conducted experiments, and discussed results at regular meetings, making the prototype of modern scientific conferences and journals. These societies succeeded the monastic tradition in a new, more secular era, continuing the mission of preserving and advancing knowledge.

Professional communities: craft guilds and trade unions. In ancient civilizations, professional knowledge was transmitted through oral tradition: craftsmen trained apprentices, passing skills from generation to generation. This practice led to the formation of craft guilds in medieval Europe, which provided training, set quality standards, regulated the labor market, and protected the interests of their members.

With the advent of the industrial age, trade unions - organizations representing the interests of workers - emerged. The first trade unions appeared in Great Britain in the late 18th and early 19th centuries as a response to difficult working conditions. They became a powerful form of workers' collaboration, allowing for collective bargaining with employers, fighting for better working conditions, higher wages, and the introduction of social guarantees.

Industrial institutions, system production, and specialization of labor: manufacturers, factories, corporations. The Industrial Revolution of the XVIII-XIX centuries radically changed the organization of labor and production, creating new forms of collaboration on a scale never seen before. Manufactories, which appeared as early as the sixteenth and seventeenth centuries, were the first step towards the systematization of production, introducing the division of labor and the specialization of workers. The factory system required the development of new methods of coordinating the actions of hundreds and sometimes thousands of workers. This led to the emergence of modern principles of management and organization of production. In the early 20th century, Frederick Taylor developed a scientific management system aimed at improving labor efficiency through its analysis and optimization.

The development of mass production technologies, especially the introduction of conveyor lines by Henry Ford in 1913, further increased the need for precise coordination and synchronization of workers' actions. This required the creation of sophisticated systems of production planning and control. As the scale of production increased and economic relations became more complex, corporations - large enterprises with a developed hierarchical management structure - emerged. Corporations such as Standard Oil and United States Steel Corporation developed management systems that coordinated the work of tens of thousands of employees in various locations.

Interdisciplinary cooperation. The 20th century brought new dimensions of collaboration, especially in the scientific field. Major scientific projects, such as the Manhattan Project during World War II, demonstrated the power of interdisciplinary collaboration. Other significant examples:

• International Space Station (1998-present)
• Large Hadron Collider to search for new subatomic particles
• Human Genome Project (1990-2003) to decipher the human genome

These historical forms of collaboration established the fundamental principles that underpin today's digital collaborations. They established the importance of sharing knowledge, coordinating efforts, and solving problems together. The transition to the digital age has not so much changed these basic principles as expanded their scope, speed, and effectiveness. From slow email exchanges, we have moved to instant global communication. From closed guilds to open information-sharing platforms, from highly specialized communities to complex ecosystems that bring together different areas of knowledge.

Digital Collaboration Mechanisms

As we can see from the excursion into history, collaboration mechanisms differ in the scale of participation (from a few people to millions), goals, and type of participants (people, machines, algorithmic systems). Collaboration processes themselves differ significantly in regulation mechanisms and quality control systems.

Today's digital collaboration arrangements are unprecedentedly diverse, encompassing a wide range of scales, actors, and purposes:

1. Scale of participation. From small teams of a few people to global projects with millions of participants. This diversity allows collaborative approaches to be applied to local initiatives and large-scale international projects.
2. Type of participants. Modern collaboration is not limited to interactions between people. It involves complex interactions between humans, machines, and algorithmic virtual systems, creating new forms of synergy between human and artificial intelligence.
3. Collaboration goals. The objectives range from solving specific business problems to global scientific research and social initiatives. The diversity of objectives reflects the universality of the collaborative approach and its applicability in different spheres of activity.
4. Mechanisms for regulation and quality control. Each type of collaboration requires a different approach to managing the process and ensuring the quality of the results. These mechanisms may include algorithmic solutions, reputation systems, peer review, and other methods.

The effectiveness of a collaborative mechanism depends on the context of its application. What works for a crowdsourcing project may not be applicable in the context of scientific cooperation or business partnership. Therefore, the choice of an appropriate collaboration mechanism requires a careful analysis of the specifics of the project, its goals, and its participants. In the table below, we have systematized various digital collaboration mechanisms, describing their key characteristics, the ideas behind them, and examples of their successful application.

The presented mechanisms illustrate the evolution of digital collaboration mechanisms. Let's look at the key stages in the evolution of digital systems.

Hypertext without the Internet (1960s-1980s). The origins of modern collaborative systems lie in the 1960s-1980s. The NLS (oN-Line System, 1968), developed by Douglas Engelbart, demonstrated the possibilities of collaborative document editing and graphical interface. The PLATO project at the University of Illinois (1960s-1970s) introduced the concepts of online forums and instant messaging that have become an integral part of modern Internet communication. The concept of living documents, described by Ted Nelsen in Literary Machines (1981), foreshadowed the emergence of modern wiki systems. These innovative ideas, however, remained mainly in the experimental and academic spheres, not becoming widespread until the advent of the World Wide Web in the late 1980s created a global platform for their practical realization and widespread use.

Asynchronous step-by-step interaction with moderation. Collaborative platforms (1990s-2000s). The creation of the World Wide Web by Tim Berners-Lee in 1989 laid the foundation for the proliferation of collaboratively edited content and the emergence and development of several revolutionary projects such as Linux (1991), Wikipedia (2001), Mozilla Firefox (2002), OpenStreetMap (2004), Arduino (2005), Khan Academy (2006), GitHub (2008), Stack Overflow (2008), and Kickstarter (2009). These platforms demonstrated the power of the collaborative approach. However, they all utilized an asynchronous collaboration mechanism followed by moderation. Despite considerable success, this approach to collaboration has faced limitations in the speed of interaction and efficiency of coordination, especially in projects that require rapid exchange of ideas and instant feedback.

Real-time collaborative tools and environments. Specialized collaboration tools (2010s). The approach based on real-time collaboration on a single document was developed in parallel. The first widely known commercial example of this approach was Google Docs, launched in 2006 based on the startup Writely, acquired by Google. The 2010s saw a trend towards specialization of collaboration tools for specific industries and tasks. Miro (2013) for visual collaboration on projects, Airtable (2015) for database management, Figma (2016) for design, Notion (2016), and Coda (2017) for workspace organization were the brightest representatives of this trend. Figma revolutionized design by enabling real-time collaboration between designers and developers. On the other hand, Notion combined note-taking, database management, and project management functions in a single platform.

Augmented collaboration: ML-Tool → ML-Agents. Today, ML agents are transforming from tools into full-fledged participants of collaborative processes, partially removing the need for large-scale human interaction to solve specific tasks. Teams consisting of humans and AI often outperform purely human teams and AI systems in solving complex problems.

Successful examples of such synergy exist in various fields: In medicine, AI helps with diagnosis, and doctors interpret results and make final decisions; in finance, AI analyzes market trends, and traders use this information to make strategic decisions; and in creative industries, AI generates ideas and options, and artists and designers select and finalize the best of them. In the next publication, we will discuss Augmented collaboration mechanisms in more detail.

Conclusion

The history of collaboration from ancient civilizations to the digital age demonstrates humanity's continuous quest for more efficient cooperation involving an ever-increasing number of interacting agents. While ancient projects, such as building the pyramids or the Great Wall of China, required unprecedented coordination of physical labor, modern digital collaborations, like Wikipedia or open source software, focus on the accumulation and organization of knowledge.

The most significant digital collaborations so far have mainly focused on systematizing existing knowledge rather than creating fundamentally new knowledge. This is reminiscent of the role of medieval monasteries and early universities as repositories and disseminators of knowledge. However, there is reason to believe we are on the threshold of a new era.

Augmented Collaboration, combining human intelligence with AI capabilities, could catalyze a new Renaissance in the digital world. Just as the invention of printing fueled an explosion of creativity and scientific discovery in the 15th and 16th centuries, the symbiosis of humans and AI opened up opportunities for creativity and innovation.

The future of collaboration lies not only in technology but also in new social and organizational models. We expect the emergence of flexible, decentralized structures where traditional hierarchies give way to dynamic networks of experts, AI agents, and distributed teams. These structures may resemble Renaissance guilds, but on a global scale and with capabilities multiplied by technology.

The key challenge will not be to develop new tools but to develop "collaborative intelligence" - effectively combining human and machine capabilities to solve the most complex problems. This will require technical skills and a deep understanding of human nature, creativity, and ethics - qualities central to the Renaissance's humanists.

Ultimately, Augmented Collaboration may be a new way of working and a fundamental shift in how we approach creativity, innovation, and solving global problems. Just as the Renaissance changed the trajectory of human civilization, the Augmented Collaboration era can open new horizons of human potential by combining the best of our historical experience with the possibilities of advanced technology.‌‌‌‌

TRIZ (Theory of Inventive Problem Solving) has a well-known law according to which the development of technical systems is aimed at increasing their degree of ideality. An ideal technical system is characterized by minimal physical parameters — its weight, volume and area tend to zero, while the ability to perform a given function is not lost. In other words, an ideal system becomes practically invisible, but retains its functionality at a high level.

In recent years, we have seen the idea of the ideal system being realized in three key areas of human-machine interface development:

• Context-aware program interfaces. Classic interfaces are being replaced by compact predictive solutions such as voice control, recommendation search and GPT chats. These technologies free the user from routine actions by utilizing context, making interaction with systems more intuitive.
• Context-aware physical interfaces. The development of large language models and high-speed internet is fostering the emergence of new ergonomic devices that enable AI-assisted applications in various domains without the need for self-input.
• Unlimited expansion of interface boundaries. Today's software and physical technologies aim to create seamless interfaces. Neurointerfaces, holography, IoT sensors, smart cities, wearable devices and augmented reality are not just making life easier, but transforming the perception of reality for both humans and machines.

By extending interfaces, we hoped to open up new possibilities for controlling machines. However, as a result, we have created a window of opportunity for machines in the human world. Machines have access to vast amounts of information, which they can process and comprehend with a depth not available to humans. As a result, they are transforming from mere tools into full-fledged actors capable not only of interpreting reality at a level approaching human perception, but also of changing it.

Controlled explosion of technological singularity

Friedrich Nietzsche, in his work Thus Spoke Zarathustra (1883-1885), introduced the idea of the superhuman (Übermensch), a being who transcended human limitations and weaknesses. Today, this vision is reflected in the concept of technological singularity — the moment when artificial intelligence (AI) will surpass human intelligence.

The concept of a technological singularity was first outlined in the mid-1960s by British mathematician Irving John Good, who in his article "Speculations Concerning the First Ultraintelligent Machine" suggested that the mental capabilities of a superintelligent machine would surpass those of humans and lead to an "intelligence explosion". Good speculated:

• "Thinking" abilities of a super-intelligent machine will surpass those of any human being, including designing even more advanced machines, resulting in an "intelligence explosion."
• A super intelligent machine will be the last invention of mankind if it can be controlled.
• The introduction of a super-intelligent machine will lead to significant changes in society and the economy, including the replacement of humans in various activities.
• One of the key tasks of super intelligent machines is to estimate the probabilities of events that have not previously occurred.

The concept of the singularity was popularized in the early 1990s in an essay by mathematician and writer Werner Vinge, "The Coming Technological Singularity: How to Survive in the Post-Human Era". Winge proposed four scenarios for achieving the singularity, each of which is being realized to some extent today:

1. Creating superintelligence through artificial intelligence. Machines will become smart enough to improve themselves.
2. Brain-computer interfaces. People will be able to connect directly to computers, greatly increasing their intellectual capabilities.
3. Biological enhancements. Advances in biotechnology will greatly increase human intelligence.
4. Network and organizational improvements. Global networks and collective intelligent systems will become so powerful that they will begin to function as superintelligence.

Vinge believed that the Singularity was inevitable due to the exponential growth of computing power and advances in AI, and suggested that this event could occur as early as the first half of the 21st century. Futurist Ray Kurzweil, in his book The Singularity is Near: When Humans Transcend Biology, suggested that the singularity will occur by 2045, when machines will become not just assistants, but equal partners, and perhaps even our teachers.

Ethical challenges and possible threats. However, there are also serious risks associated with these prospects. For example, Nick Bostrom, in his TED talk "What happens when our computers get smarter than we are?" warns that uncontrolled use of AI could lead to catastrophic consequences if AI decides to transform the entire planet to maximize the efficiency of the task at hand.

It is important not only to limit choices, but also to correctly envision and clearly define goals for AI. We can anticipate threats and plan workarounds, but we cannot shut down a system on which we ourselves depend, or if the system can "shut us down".

AI can be placed in a virtual simulation of reality, but there is no assurance that it will not find a way around this isolation. It is important to ensure that AI, if it leaves the virtual environment, remains on the side of humanity and shares our values. These values must be considered not only in known contexts, but also in uncertain futures where new ethical dilemmas may arise.

Solomon's ring: controlling the djinn

In the legends of King Solomon, the ring gave the king power over demons and supernatural beings, providing wisdom and control over forces that were beyond the reach of ordinary people. In today's world, this image is reflected in the physical component of new types of interfaces.

Interfaces will not require training; rather, they will train users to interact better to achieve the desired result. Learning and honing the skill of blind typing techniques on a keyboard takes hundreds of hours. Users learn key placement and train themselves to type quickly and accurately without having to look at the keyboard. Similarly, switching to a new program or operating system was a barrier to using a variety of programs.

AI advances have resulted in the development of intuitive and adaptive interfaces. Today, voice assistants Siri, Amazon Alexa and Google Assistant allow users to interact with technology using natural language. Users don't need to learn commands or menus — just ask a question or give a command.

Interfaces will become fully personalized. The execution of tasks by machines will be driven by context rather than specific requests, leading to truly personalized assistants. Research McKinsey & Company и Salesforce shows that about 70% of consumers expect personalized interactions from companies, and about 70% of them are disappointed when it doesn't happen. These expectations are driving companies to invest in technology that can more accurately anticipate user needs and deliver relevant solutions. In fact, personalization has already become the new norm.

Alexa and Google Assistant voice assistants are already adapting to user requests based on previous interactions and the current situation. In the future, such systems will become contextually aware, taking into account the user's schedule, habits, preferences and emotional state. Analysis of behavior and preferences will allow interfaces to adapt to needs and anticipate actions.

Interfaces will more actively use multimodal approach and extra-textual ways of interaction. Multimodal interfaces will combine several channels of information input and output: biometric state, body position, gestures, gaze, facial expressions, voice, including its emotional coloring. An example of such technology is Apple Vision Pro, which allows you to control various device functions using hand gestures and eye movements.

New interfaces will make greater use of non-verbal and low-energy interaction methods such as gestures, facial expressions, and biometrics to perform tasks without explicit commands from the user. The data will not be used in isolation, but in combination, creating an overall impression of the user's state. Technologies that recognize and respond to emotional state will combine speech and facial expression analysis for a more accurate and contextualized response. Examples of projects developing these ideas:

• Microsoft Azure Kinect is a platform for building applications that recognize motion and analyze biometric data, used in healthcare, manufacturing and education.
• Microsoft HoloLens allows you to interact with holographic objects in the real world.
• Google Soli uses compact radars to recognize gestures, allowing you to control devices without physical contact.
• Affectiva and Emotient (acquired by Apple) analyze facial expressions and voice tone for real-time emotion recognition. The solutions are used in automobiles to improve safety, retail and marketing to gauge consumer reactions.

The ergonomics of AI-assisted devices will change to more effectively gather information about the user's condition. In the 20s, many innovative devices have hit the market, each offering new ways to interact with the user and the world around them:

• Tab AI. A small disk is worn around the neck and continuously records speech. The recordings are sent to the cloud and processed by AI to transcribe the speech and provide analytics.
• Rabbit R1. A compact device with a touchscreen and rotating camera. The device acts as a personal voice assistant to perform AI-assisted tasks from booking a cab to finding recipes via voice commands.
• Humane AI Pin. A small square device with a gesture recognition camera and a miniature projector attaches to clothing with a magnet. It is controlled by voice and gestures.
• Ray-Ban Meta Smart Glasses. Sunglasses with built-in audio and video features to take photos, record videos, listen to music and take phone calls using voice commands and a touchpad on the temple.

The output of visual information will not require physical displays. Ericsson, for example, has developed a system holographic communication systemwhich allows to create three-dimensional images using ordinary smartphones or tablets. It is also worth noting the products of ARHT Media, which offers solutions for educational institutions and corporate clients, using 4K holographic touch displays to create realistic three-dimensional holograms.

The Delphic Oracle: cooperation with the supermind

Transparency of the interface will be ensured by its asymmetry: wide-channel systems of continuous data collection and compact intuitive decision-making systems. In the concept of collective intelligence, computers process huge amounts of data and propose solutions, while people, relying on their intuition and experience, make the final decisions.

Wide-channel interfaces such as sensors and data analyzers will collect and process vast amounts of information. These systems will function in the background, with little or no direct interaction with the user, but will provide important information for decision-making. For example, in medicine it is the collection of biometric indicators and test results, and in city management - all information about traffic, energy consumption and public safety.

In turn, decision interfaces will be as compact as possible. Sarah Gibbons and Kate Moran of Nielsen Norman Group argue that future interfaces will combine conversational AI with microinterfaces — small, specific interface elements that unobtrusively give users exactly the information and control they need at that moment.

Not only information processing tasks, but also selection tasks will be taken over by machines. With the growth of computing power and the improvement of machine learning algorithms, AI systems will be able not only to collect and analyze data, but also to make complex choices based on it. This means that machines will be able to make decisions in situations that require the evaluation of multiple factors and a quick response:

• Autonomous driving. Already today, Tesla cars with Autopilot make decisions in real time by driving vehicles. Future systems will be able to assess road conditions, predict the behavior of other road users and choose optimal routes with minimal human intervention. In logistics, autonomous trucks like those being developed by Waymo can optimize routes and reduce operating costs.
• Medicine. AI will be able not only to diagnose diseases, but also to choose optimal treatment methods based on the analysis of medical data, genetic information and the latest research conducted, including by the artificial intelligence itself.
• Emergencies. AI systems will play a key role in dealing with emergencies such as natural and man-made disasters. AI will be able to quickly analyze data from drones and satellites, identify the most affected areas and dispatch rescue teams. The systems will also be able to predict the development of emergencies and take measures to prevent them, minimizing damage and losses.

In the future, manual data entry and retrieval will no longer be necessary. Users will set the direction of the system and validate the results. Tools will become describing and explaining mechanisms, allowing people to focus on strategic management and decision making. The high level of trust in AI assistants will lead users to delegate even decision-making tasks to machines. This poses possible dangers associated with over-reliance on AI, including risks of loss of control.

Biotechnological fusion will become possible, which will expand human capabilities and improve the quality of life. Current developments in biotechnology and AI are opening up opportunities for the fusion of biological and technical components that will greatly expand human capabilities and improve quality of life. The article "The Future of Bionics: Prosthetics and Neural Interfaces" (MIT Technology Review, 2021) describes bionic prosthetics that integrate with the user's nervous system, allowing control of movement through thoughts. Research in neural implants, such as described in "Memory Implants: The Next Frontier in Neurotechnology" (Scientific American, 2020), shows promise for improving memory and cognitive abilities. Genetic engineering, discussed in the article "CRISPR and Beyond: The Future of Gene Editing" (Nature, 2019), offers opportunities to change human abilities and improve health.

• Neuralink. Ilon Musk's company that develops implantable neural chips to create brain-computer interfaces. These chips allow users to control computers and other devices with their thoughts.
• Synchron. An implantable device that can transmit brain signals to a computer and does not require open brain surgery.
• DARPA's RAM. A project to create implantable devices to restore memory in soldiers with brain injuries.
• CTRL-Labs (Facebook). The CTRL-Kit bracelet is able to recognize electrical signals from the nerve endings of the hand, allowing you to control gadgets and PCs with minimal finger movements.
• Paradromics. High-speed brain-computer interfaces that help people with severe physical disabilities regain some functions.

Conclusion

The future of human-machine interfaces is an inevitable step towards better integration of technology and everyday life. We are already seeing interfaces become more intuitive, context-aware, and virtually invisible. New interface solutions are blurring the boundaries between man and machine, making technology a natural extension of our capabilities.

However, along with these achievements come serious challenges. The technological singularity, which many futurologists believe is just around the corner, not only opens up new horizons, but also presents risks. It is important to lay down ethical standards and control mechanisms for AI in advance, so that it serves for the benefit of humanity and does not become a threat.

The interfaces of the future, whether neural interfaces or augmented reality, may become our indispensable assistants. But along with empowerment, we will have to take responsibility for their development and use to ensure that technology serves society rather than dictating new rules.

The 2000s marked the beginning of a new era in technology. Technology has become not only intuitive but also personalized. The massive proliferation of smartphones has been the catalyst for this progress.

On the one hand, smartphones have given companies unprecedented access to user data and an in-depth understanding of each individual's needs. On the other hand, smartphones have democratized access to computing power and the Internet, making technology available to significantly more people.

Below, we explore the key moments and milestones that shaped the development of human-machine interfaces in the first two decades of the 21st century.

New concepts for machine-human interaction

1. Virtual worlds are replacing reality. With the spread of the Internet and the development of cloud technology, it is now possible to live in virtual worlds with different levels of engagement. These range from social networks with daily updated digital profiles to purpose-built game worlds using Oculus Rift and other VR devices that allow immersion in entirely virtual environments where users can experience adventure, work, and learn.

Virtual spaces are becoming so integrated into everyday life that the boundaries between the physical and digital worlds are beginning to blur. We can spend time with friends on social media and then switch to work projects without feeling the transition between two different worlds. Virtual worlds represent a new stage in the evolution of human-machine interfaces, where our digital lives become as meaningful as our physical ones.

2. Smartphone as a key meta-interface. The smartphone or similar device becomes a pivotal meta-interface to the digital world. These devices provide access to the Internet and various applications and serve as central hubs for managing digital lives. Smartphones integrate multiple functions, from communications to entertainment to smart home management, making them universal interfaces for interacting with the digital world.

Biometric authentication mechanisms, such as Apple's Touch ID or Face ID, and crypto interfaces, are emerging to ensure data security and privacy. These technologies provide secure and convenient access to personal data and services, protecting users from unauthorized access. Smartphones thus become not just devices but critical elements that connect us to our digital avatar.

3. Blurring interface boundaries through intelligent systems. Interface boundaries are beginning to blur through voice control and systems with smart search. Voice assistants such as Siri, Alexa, and Google Assistant allow users to interact with technology using natural language, making the process more intuitive and convenient.

These systems execute commands and adapt to user preferences, considering their previous actions and surrounding context.

Intelligent systems can guess our wishes and offer solutions before we know our needs. For example, smart homes can automatically adjust lighting and temperature according to the time of day and our habits. At the same time, digital assistants can suggest optimal routes or remind us of upcoming appointments.

In addition, commercial tools are emerging that can convert voice, read gestures, convert text from images, and perform other complex tasks. These technologies, such as gesture recognition and OCR (Optical Character Recognition) systems, greatly enhance machine interaction by making interfaces even more invisible and intuitive.

New HCI technologies and devices

Devices with multi-touch input: smartphones and tablets. The early 2000s were marked by the active adoption of multitouch technology, which allowed users to interact with devices using multiple touchpoints simultaneously. The release of the first iPhone in 2007 was revolutionary in the cell phone industry. Users were offered a touchscreen-based interface that radically changed the mechanics of interacting with mobile devices.

The release of the first iPhone in 2007 was revolutionary in the cell phone industry. Users were offered an interface that was entirely touchscreen-based, which radically changed the mechanics of how users interacted with mobile devices.

Wearable devices such as fitness trackers (e.g., Fitbit) and smartwatches (e.g., Apple Watch) that track physical activity and health emerged in the mid-2010s. These devices offer convenient ways to interact with digital assistants and notifications, help users track their health, and provide data for physicians to analyze.

NFC payments. By the end of the 2000s, the first NFC-enabled devices appeared, which made it possible to use smartphones for contactless payments in stores, making the process more convenient and faster.

Internet of Things (IoT) and Smart Home. Connecting various devices into a single network allowed them to communicate with each other and users via the Internet. IoT has widespread use in smart homes, industry, and healthcare, improving automation and control over various processes.

Smart speakers. Devices such as the Amazon Echo (2014) with Alexa voice assistant, Google Home (2016), and Apple HomePod (2018) have given users the ability to interact with digital assistants through voice commands to control a smart home, play music, get information, and more.

AR/VR glasses Google Glass (2013) and Oculus Rift (2016) showed the potential of AR and VR technology. Oculus Rift became the technology that defined the direction of modern VR helmets and contributed to the revival of interest in virtual reality in the gaming industry and other fields. VR and AR began to be used extensively for simulation training in fields such as medicine, aviation, and military, as well as for visualizing projects in engineering and architecture. This allowed users to interact with virtual objects more intuitively using natural hand and body movements.

In the mid-2010s, with the development of motion sensors and improved gesture recognition algorithms, these technologies began integrating with VR/AR systems. Such systems allowed users to interact with virtual objects more intuitively using natural hand and body movements.

Gesture control devices. Microsoft's Kinect (2010) and Leap Motion (2012) gesture control devices showed the potential of gesture control for games and professional software, offering new ways to interact with digital content. Kinect uses advanced cameras and depth sensors to control games in real-time through user movements and gestures. These technologies have applications in various fields, including music production, graphic design, and virtual reality.

Autopilot cars. The Google Self-Driving Car and Tesla Autopilot projects, which began in the late 2000s and early 2010s, used sensors and artificial intelligence to control vehicles without driver input. These technologies promise to revolutionize the transportation industry by improving the safety and efficiency of driving.

3D printers. The development of affordable 3D printing in the 2000s and 2010s allowed many users to translate their ideas into physical reality. MakerBot's Replicator, first released in 2012, significantly contributed to the popularization and spread of 3D printing technology. The printer became available to many consumers, including educational institutions, designers, engineers, and hobby enthusiasts.

In the medical field, 3D printing has been used to create customizable prosthetics and medical instruments, in education to produce didactic materials and training models, and in everyday life to create personalized user interfaces and home automation control devices.

Exoskeletons and other variations of Human Augmentation. Technologies such as exoskeletons often use microprocessors to process sensor data and control movements. These devices help users perform physical tasks by increasing their strength and endurance and have applications in medicine, industry, and the military.

New HCI software solutions

Social networking. The proliferation of high-speed Internet and personal computers in the early 2000s led to the emergence of social networking platforms such as MySpace (2003) and Facebook (2004). These platforms allowed users to keep in touch with friends and family and laid the foundation for a global communication network. In this network, a personalized profile is an example of a digital meta-interface that can be used on various devices - a personal computer, cell phone, or tablet.

Internet speeds have become sufficient for data to be stored in the cloud and downloaded as needed, reflecting the trend toward universal interaction systems. However, this has also led to the need to protect personal data, as in the GDPR in the European Union (2016) and CCPA in California (2020).

Messengers. With the development of social networks, messengers—instant messaging programs—developed in parallel. Some of the first were ICQ (1996), AOL Instant Messenger (1997), and MSN Messenger (1999). In the 2000s and 2010s, new messengers such as WhatsApp (2009), Telegram (2013), and Facebook Messenger (2011) emerged, which provided users with the ability to communicate in real-time via text messages, voice calls, and video calls.

Messengers have become integral to digital communication, offering users fast and convenient tools for personal and business communications. In 2016, Facebook announced support for chatbots in Messenger, which opened a new page in the history of interactive communication. Companies could now interact with customers, run marketing campaigns, and provide support through autonomous bots.

Chatbots in social networks and messengers. In 2016, Facebook announced support for chatbots in Messenger, which opened a new page in the history of interactive communication. Companies could now interact with customers, run marketing campaigns, and provide support through autonomous bots. Other platforms, such as Telegram, Slack, and WhatsApp, have joined the trend by offering tools for developing and integrating chatbots. The development of advanced language models, such as OpenAI's GPT in the late 2010s and early 2020s, has dramatically expanded the capabilities of chatbots, making them capable of more complex and contextually relevant communication.

Mobile operating systems: iOS and Android. The first version of the operating system for Apple smartphones was introduced along with the device itself. Google officially announced the Android OS in November 2007. The first commercial Android-based smartphone, the HTC Dream, was released in October 2008. Android quickly gained popularity due to its openness and adaptability, which allowed many manufacturers to create a variety of devices on this platform.

Siri and Alexa voice assistants. The logical continuation of the chatbot concept was the idea of a voice assistant. In 2011, Apple introduced Siri, the first voice assistant that became integral to the iOS ecosystem. In 2014, Amazon introduced Alexa, which became the centerpiece of the Echo smart speakers. Alexa quickly became famous for its ability to control smart home devices and provide user interaction through natural language.

Integrating voice assistants with the IoT has opened up new possibilities for smart home control. Voice commands allow users to interact with various devices—from lights and thermostats to large appliances—using voice commands.

Payment apps Google Wallet and Apple Pay. Google introduced Google Wallet in 2011, and Apple launched Apple Pay in 2014, allowing users to purchase their smartphones as digital wallets. These technologies have made it possible to integrate payment functions directly into mobile apps, simplifying the shopping process and expanding the possibilities for mobile commerce.

Blockchain and cryptocurrencies. In 2008, Satoshi Nakamoto published a white paper describing the concept of the first cryptocurrency, bitcoin. In 2009, the Bitcoin network was launched and became the first decentralized digital currency based on blockchain technology. This technology allows for secure and transparent transactions without intermediary banks. In the 2010s, blockchain was found to be applicable not only in the financial sphere but also in other areas, such as logistics, voting, smart contracts, and digital identification.

Biometric authentication systems. Biometric authentication technologies, such as Apple's Touch ID (2013) and Face ID (2017), which use fingerprints and facial recognition, have greatly improved the security and convenience of accessing devices and services. Voice authentication has also become essential, offering more convenient and secure ways to log into systems and applications.

Remote working technologies. Tools and platforms like Zoom, Slack, and Microsoft Teams have made remote working and collaboration more accessible by offering video conferencing, messaging, and real-time document sharing. These technologies became especially relevant in 2020 when the COVID-19 pandemic forced many companies to adopt remote working.

Conclusion

The 2000s-2010s marked a significant shift in the development of human-machine interfaces, making technology more intuitive and accessible. Smartphones and touchscreens, social media and messengers, virtual reality, and 3D printing — all these innovations have empowered users and deepened their interaction with digital environments.

With the development of voice assistants and smart home systems, users have begun to manage technology more easily and naturally. Biometric authentication, blockchain technologies, and payment applications have improved the security of digital services.

The past decades have laid the groundwork for future interface development, focusing on deepening the integration of artificial intelligence and machine learning into everyday life.

The 1980s laid the foundations for many of the technologies we use today. In addition to the development of voice technology, touch screens, and the mass proliferation of personal computers, this period was marked by several significant innovations in user interfaces.

New concepts for human-machine interaction

1. Compactness and mobility are as important as computing capabilities. Technical advances in microelectronics, processor miniaturization, efficient batteries, and other components have made it possible to compact computers and combine their computing capabilities with other functions - primarily telecommunications and multimedia data.

During this period, new portable devices were created: pagers, cell phones, communicators, laptops, the first e-books, MP3 players, and digital cameras. These devices allowed users to stay connected and work on the move. These devices changed how we communicate and work and empowered people to accomplish professional and personal tasks from anywhere in the world.

The social significance of the changes has also been enormous: People have been empowered to work and socialize, but at the same time, they have become more dependent on constant connectivity. Constant accessibility and engagement in communication have created new dynamics in work and personal life, blurring the boundaries between them and creating new challenges for time and attention management.

2. The computer as a tool for creative tasks. In the 1980s and 1990s, computers became a tool for utilitarian tasks and a powerful medium for creating and consuming content in art, design, music, and film. Technological advances in graphical interfaces, increased computing power, and specialized software development opened new horizons for creative professionals.

The advent of graphic design and digital music creation programs has allowed artists and musicians to create high-quality work directly on the computer. Video editors and modeling have revolutionized the editing of movies and videos, creating complex visual effects and animations.

3. The computer as a means of consuming content. In addition to creating content, computers have significantly changed how content is consumed. The advent of multimedia PCs has allowed users to watch movies, listen to music, and view photos on a single device. These technological advances have expanded opportunities for creative expression and made art and media more accessible to a wider audience, changing the landscape of entertainment and culture.

Interaction with computers has allowed users to act in different roles. Modern computers made it possible for a person to be both a creator and a consumer. The user could program a website, record songs, edit photos, and then switch to watching movies or listening to music, taking on different roles depending on his or her needs and mood.

New HCI technologies and devices

Pager. The first pagers appeared in the 1950s and were used in hospitals and emergency services. In the 1980s, they became popular in business and personal communications, becoming more compact and functional. Pagers had monochrome displays for text messages, notified users with beeps or vibrations, and allowed messages to be viewed with a few buttons to scroll, delete, or confirm. Most pagers could only receive messages.

Cell phone. In 1983, the Motorola DynaTAC 8000X, the first commercially available portable cell phone, was released. The cell phone era began, allowing people to communicate on the move and outside traditional work or home spaces.

In 1994, IBM launched the IBM Simon, the first smartphone prototype that combined the functions of a cell phone, communicator, pager, fax, and e-mail device. In addition, Simon became one of the first commercially successful touchscreen phones. The device cost about 1000 dollars. Subsequently, these technologies continued to evolve and became the basis for modern text input methods on touch screens and virtual keyboards.

Communicator. In the 1990s, other wearable devices, such as communicators and personal digital assistants (PDAs), became popular, contributing to the concept of the mobile office. Such devices could already combine the functions of a cell phone and a device for accessing e-mail and could synchronize with a personal computer.

The stylus helped users interact more accurately and comfortably with touchscreens, especially when typing and controlling small interface elements. The technology was later used in miniature phones like Nokia and Samsung series devices. Today, some electronic pens in tablets and modern smartphones inherit this idea, allowing users to enter data and draw on the screen.

CD-ROM. The development and proliferation of CD-ROMs in the 1980s significantly impacted multimedia technology, providing a new way of storing and playing back audio, video, and text information, which influenced the development of interactive interfaces. This technology increased the amount of data available for processing and storage, which played a key role in developing programs such as Encarta and Adobe Photoshop, as well as the first computer games with rich graphics and sound, such as Myst and The 7th Guest. These games were among the first to exploit the full potential of CD-ROM, offering deeply immersive virtual worlds with high-quality visual and audio effects that were previously unavailable due to the limitations of earlier media. CD-ROMs also spurred the creation and distribution of multimedia training programs, allowing educational institutions and companies to offer more interactive and engaging training materials.

Laptops. Portable computers such as the IBM ThinkPad and Apple PowerBook made computing power available outside the office or cubicle.  With improvements in component and battery power efficiency and the advent of smaller hard disks and LCD screens, laptops became increasingly lighter and thinner. These innovations allowed users to perform complex tasks and stay connected while on the go, laying the groundwork for today's concepts of remote working and digital nomadism.

E-book. E-book readers such as the Rocket eBook (1998) allowed users to carry entire libraries, making reading more convenient and accessible. These devices utilized new electronic ink (e-ink) technology that provided high contrast and low power consumption, making reading comfortable and long-lasting

Digital cameras. Mass-market digital cameras began to appear on the market in the late 1980s and early 1990s. One of the first mass-market digital cameras was the Dycam Model 1, the Logitech Fotoman. Subsequently, many companies made their own versions of the digital camera. Apple even had its own - the QuickTake 100. Early models had a resolution of tenths and hundredths of a megapixel and could only store a few dozen images, but they did so instantly and without the need to develop film.

Portable game consoles. Portable game consoles, such as the Nintendo Game Boy (1989) and Sega Game Gear (1990), revolutionized the world of video games by allowing you to play your favorite games anywhere, anytime. The devices were equipped with a screen, a built-in controller, and a rechargeable battery, allowing gamers to enjoy the game regardless of the presence of a television or power outlet. With its simple monochrome graphics and long battery life, the Game Boy gained popularity for its many games, including the famous Tetris. On the other hand, the Sega Game Gear offered a color screen and more powerful features, attracting the attention of game enthusiasts with its high-quality visuals and variety of gaming options.

The first virtual reality helmets. Although VR had not yet become a mainstream phenomenon in the 90s, this period was rich in research and development in the field. In 1991, Sega announced the Sega VR, one of the first attempts at a VR system for the consumer market. These same years also saw the first attempts at gesture interfaces. Oblong in 1992 offered a system that allowed virtual objects to be controlled using gestures.

GPS navigator. The first GPS navigators were developed for military use in the 1960s. In the 1990s, civilian GPS units became compact and available to the general public, especially in automobiles. The devices had monochrome displays that showed real-time maps and coordinates, allowing users to enter addresses to plot a route. GPS navigators provided turn-by-turn directions, showing speed, direction, and time to arrival, greatly improving the road's convenience and safety.

New HCI software solutions

T9. As more and more input options became available, more engineers tried to make the input process more convenient and faster. In 1995, Tegic Communications introduced T9 (Text on 9 keys), a predictive text input technology for cell phones. T9 allowed faster text input on devices with a keyboard where each key corresponds to several letters (as on a traditional cell phone keyboard). T9 used a dictionary and algorithms to predict the word the user intended to type, making text input much easier and faster.

MacOS and Windows. One of the key achievements of the late 1970s was the creation of the graphical user interface (GUI), which revolutionized the way people interacted with computers. The Xerox Star, introduced in 1981, was the first commercial product with a GUI, offering an intuitive way to work through graphical images, windows, icons, and menus. Apple built on this initiative with the release of the Macintosh in 1984, which was the first mass-market PC with a GUI, making the technology accessible to a wide range of users.

The competition between Windows and Mac OS in the 1980s and 1990s played a key role in the evolution of user interfaces. In 1984, Apple released System 1, offering a revolutionary graphical interface with windows, icons, and a shopping cart. In 1985, Microsoft introduced Windows 1.0, the first graphical user interface on top of MS-DOS. In 1990, Microsoft released Windows 3.0 with improved interface, performance, and multimedia support, making it a serious competitor to Mac OS. Apple responded in 1991 with the release of System 7, which introduced color interfaces, support for virtual memory, and improved multitasking capabilities. The competition continued in 1995, when Microsoft introduced Windows 95, with the Start button, taskbar, and menu system that became the standard for subsequent versions.

In the late 1980s, Apple released its first Human Interface Guidelines, which had a profound impact on user interface design. These guidelines helped developers design intuitive, efficient, and visually appealing interfaces for software running on the Macintosh. The guidelines articulated key concepts such as consistency, direct manipulation of objects, user feedback, the visual metaphor of the desktop, and user control. These principles contributed to the unification of Macintosh application interfaces and significantly impacted interface design principles in general, shaping approaches still in use and evolving today. Similar guidelines were produced for Microsoft Windows and other systems such as IBM and Sun Microsystems. In the early 1990s, the idea also entered the Open Source community, creating unified interfaces in the Motif GUI for Unix.

Linux. In the 1980s and 1990s, Linux offered an alternative to commercial operating systems such as Windows and Mac OS with terminal and graphical solutions. Based on free software and open-source code, Linux gave users flexibility and control that proprietary systems did not have. In the 1990s, the first graphical environments for Linux appeared, such as the X Window System, KDE (1996) and GNOME (1997). These environments provided functionality comparable to commercial GUIs but with more customizability and without license restrictions. In addition, Linux was less demanding on resources, allowing it to run on more devices and to be used efficiently for the Internet and servers, making it attractive to developers and enthusiasts looking for an alternative to commercial operating systems.

Graphics programs for working with multimedia. Photoshop, released in 1988, allowed artists and designers to edit and process images on an unprecedented scale, ushering in an era of digital art. Pro Tools, introduced in 1991, radically changed the sound recording process by providing musicians and sound engineers with powerful tools for multitrack recording and digital audio editing. Such programs expanded the creative possibilities of professionals and became catalysts for entire industries, providing the technological foundation for modern media content creation and distribution methods.

WWW.  1989 Tim Berners-Lee proposed the World Wide Web (WWW) project, which in the early 1990s led to the first websites and browsers such as Mosaic and Netscape Navigator. By the late 1990s, this led to the massive spread of the Internet and the development of web interfaces, making information available to millions of users worldwide. Web interfaces became intuitive and functional, making navigating and interacting with Internet resources much easier. This project has transformed how we communicate, work, and receive information, profoundly impacting society and the economy.

The first WWW server based on the NeXT computer at the CERN Museum

HTML and browsers. HTML (HyperText Markup Language) was first developed by Tim Berners-Lee in 1991. The first web browser, WorldWideWeb, also created by Berners-Lee, appeared in 1991. In 1993, the Mosaic browser was released, the first graphical interface browser to facilitate the mass adoption of the Internet. Then, in 1994, Netscape Navigator was released, quickly becoming popular for its advanced features and ease of use. These technologies played a key role in forming and developing the World Wide Web, making the Internet accessible and convenient for many users.

Search engines. The launch of Altavista in 1995 and Google in 1998 radically changed the way people search for information on the Internet, making it more accessible and user-friendly. In many modern tools, part of the dialog system is analogous to the smart search bar. In addition, the first systems for digital communication are beginning to appear. For example, in 1996, the ICQ instant messaging program became available.

Continuous speech recognition programs began to gain traction in 1986/ when Carnegie Mellon University released Sphinx, a system capable of recognizing spontaneous speech using hidden Markov models, an advanced approach in natural language processing. In the early 1990s, Dragon Systems released Dragon Dictate, the world's first commercially available voice recognition system for the general public, which allowed users to dictate text to a computer but required pauses between words for more accurate recognition. In 1997, the same company released Dragon NaturallySpeaking, the first continuous speech recognition system that allowed users to speak naturally and without pauses. The product used sophisticated algorithms and machine learning technology to adapt the system's behavior to the user's voice and accent.

Conclusion

The history of human-machine interfaces in the 1980s and 90s demonstrates how technological innovations transformed how people interact with computers. These decades were packed with significant advances in graphical interfaces, mobile and compact devices, and creative tools that allowed users to be consumers and content creators. The development of technologies such as graphical user interfaces (GUIs), speech recognition systems, and the World Wide Web (WWW) have played a key role in shaping today's digital landscape.

The changes have created new dynamics in people's working and personal lives, making communication and access to information faster and more convenient. However, they have also created new challenges, such as dependence on constant communication and the need to manage time and attention.

In the 1960s-70s, engineers redefined computers' role and capabilities in everyday life, creating the first graphical interfaces and multi-user systems. Computers began to take on new forms: alongside specialized computing devices, there emerged personal machines, game consoles, and programmable robots.

The diversity in the forms of computers required a reassessment of approaches to human-machine interaction, leading to the creation of new physical and software interfaces. Initially, users interacted with the operating system through a command line. Later, graphical programs were developed, where people controlled the behavior of objects by moving a mouse cursor.

Below, we will examine the key concepts of human-machine interaction that emerged in the 1960s-70s, the devices and software solutions inspired by these ideas, and the developments of previous years.

New Concepts of Human-Machine Interaction

Advances in microprocessors during the 1960s led to significant miniaturization and increased transistor density. This allowed for the design of compact computers, enhancing human capabilities in storing and processing information.

In the 1960s-70s, engineers proposed three new approaches to understanding the future development of human-computer interaction:

1. The computer is a universal tool for professional work.
2. A network of computers provides emergent properties that cannot be achieved by individual machines alone.
3. The primary vector of HCI progress is virtual reality, where the interface becomes invisible.

All these ideas aimed to make computers more accessible to humanity, not just to niche engineers. Below, we will analyze the premises of these ideas, their implementation features, and physical and software artifacts.

1. The Computer as a Universal Tool for Professional Work: In the 1960s-70s, engineers realized that computers could solve computational tasks and be effective tools in most areas of human activity.

For example, in 1967, on a CBS show, Walter Cronkite presented the concept of the home office of the future. The show featured a phone with a video screen and consoles for accessing news and weather, anticipating the modern use of separate applications for various tasks.

In December 1968, a team led by Douglas Engelbart conducted a demonstration known today as The Mother of All Demos. The presentation showcased the following innovations: video conferencing, hypertext, and collaborative work on documents and files. The demonstration set the direction for the further development of computer interfaces, their clarity, intuitive understanding, and flexibility.

2 Computer Networks. Launched in 1969, ARPANET became the first successful example of packet switching for long-distance communication. ARPANET connected several universities and research centers and demonstrated the possibilities of remote data exchange and collaboration. In the early 1970s, the network included about 20 nodes, and by the end of the decade, the number had increased tenfold. In 1973, the NORSAR network was launched in Norway, becoming the first network connected to ARPANET outside the United States.

ARPANET served as a platform for information exchange between institutions and a testing ground for engineers to develop and test new networking technologies and protocols. Scientists used email for information exchange and the FTP protocol for file sharing. In 1973, the first experiments in voice transmission over the network were conducted, and in 1978, the TCP/IP protocol was tested. By the late 1970s, work began creating the Domain Name System (DNS), which simplified navigation and network management. These achievements laid the groundwork for creating the global Internet, which would change the world in the following decades.

3. Virtual Reality. In 1968, engineer Ivan Sutherland proposed the concept of The Ultimate Display. This project anticipated the development of virtual and augmented reality technologies, combining immersive experiences with visual, auditory, and tactile perception, computer modeling of physical laws, and interactivity, allowing users to create and modify virtual objects.

The military was the first to recognize the potential of virtual reality. In the 1970s, software simulators were actively used for training pilots and tank operators. For instance, the SIMNET program allowed for the creation of realistic combat scenarios and improved soldier training.

The Videoplace project (1974) by Myron Krueger enabled users to interact with virtual objects and each other in real-time. A system of cameras and screens with virtual reflections was used to track user movements. Similarly, in the 1970s, prototypes of virtual reality systems began to be used in medicine to train surgeons. These systems allowed doctors to practice complex operations in a virtual environment, reducing risks and improving skills.

In 1978, the Aspen Movie Map project, sponsored by DARPA, was presented at MIT. Users could virtually navigate the city of Aspen using a specially transformed series of photographs. This was one of the first virtual reality applications for real-world navigation, similar to how we use applications like Google Maps today.

The further development of technologies aimed to ensure a scenario where computers could simulate physical processes indistinguishably from reality. It became clear that any HC interface within this paradigm would inevitably become more compact, simpler, and universal until it became completely invisible.

New HCI Technologies and Devices

In the 1960s and 1970s, new input devices were proposed: the light pen, mouse, touch screen, and joystick. Keyboards became standardized. The personal computer almost took on its modern form, dividing into separate components. New ways of integrating computers into everyday life, such as game consoles and robots with built-in microprocessors and sensors for the external world, emerged.

Light Pen and Sketchpad. In 1963, Ivan Sutherland created the first computer drawing program, Sketchpad. The program used a new device — the light pen, which allowed direct interaction with the computer by drawing and modifying graphic objects on the screen.

Sketchpad profoundly impacted the development of graphical user interfaces (GUI) and computer-aided design (CAD) systems.

Stylus. Early experiments with styluses for interacting with electronic devices date back to the late 1950s-1960s. One of the first devices using a stylus for data input was the RAND Tablet, also known as Grafacon. The device was used to draw and input information into a computer. In the 1960s, Bell Labs created the Stylator, which allowed handwritten text to be entered into a computer and even partially recognized..

Computer Mouse. In the 1960s, Douglas Engelbart developed the concept of augmenting human intellect. This concept aimed to create interfaces that made computer interaction more intuitive and efficient. One of the outcomes of this work was the first computer mouse (1964). Engelbart called it the “X-Y position indicator for a display system.” The device consisted of a wooden case with two perpendicular wheels inside. The mouse was first demonstrated at The Mother of All Demos in 1968. Despite the demonstration's success, the mouse remained an experimental device for several years.

The first commercially available computer mouse was released in 1981 by Xerox alongside the Xerox 8010 Star Information System. The device was developed based on Engelbart’s original concept, refined by engineers at Xerox PARC. The mouse featured a more ergonomic design and used a ball mechanism for tracking movement, making it significantly more user-friendly.

Resistive Touch Screen. In 1967, researchers from CRES created the first touch screen that responded to touches. Engineer Sam Hurst improved the technology in 1972, creating the first mass-produced touch terminal, Elographics, for controlling computer systems. These innovations greatly enhanced the interactivity and usability of computers, paving the way for developing more sophisticated touch devices and applying computers in environments where mechanical keyboards were impractical or cumbersome.

Graphical Terminal. In the 1960s, text terminals and teletypes, previously used for communicating with machines, began to be replaced by devices with cathode ray tubes (CRT) and microprocessors. These new graphical terminals allowed multiple input-output devices to be connected to a single powerful computer simultaneously, significantly expanding user interaction capabilities.

Personal Computer. Developments in the 1960s led computers to acquire interfaces accessible to specialists and the general public. While traditional large computer systems continued to serve large-scale computational operations, personal computers with graphical interfaces were aimed at individual use by technically less savvy people.

In 1973, Xerox introduced the Alto, one of the first personal computers with a graphical interface. It included a windowing system, a mouse, and capabilities for working with text and graphics. This was a significant step forward compared to the previous command-line interfaces and laid the groundwork for future graphical interfaces such as the Apple Macintosh and Microsoft Windows.

Game Console, Cartridge, Joystick. In the 1970s, game consoles played a significant role in personalizing interaction with technology. Before their advent, gaming devices were bulky arcade machines installed in public places.

One of the first significant achievements was the 1977 release of the Atari 2600. The console completely transformed the industry with its modular system, allowing players to purchase and swap game cartridges and use intuitive interaction mechanisms—joysticks and paddle controllers.

Robot with an Internal Computer. With the advent of miniature computers embedded directly into robots' bodies, the mechanics of human-machine interaction in industrial settings changed significantly. The first industrial robot, Unimate, installed in 1961 at the General Motors factory, could perform tasks requiring precision and repeatability, such as part handling and welding, thanks to microprocessor solutions.

The development of sensors and actuators during this period also significantly improved robots’ perception of their surroundings and ability to perform more complex physical interactions. The Shakey project (1966), implemented at the SRI International Research Center, became one of the first autonomous robots capable of analyzing their physical environment and making independent decisions based on the information received.

New HCI Software Solutions

With the development of new concepts of human-machine interaction and the emergence of innovative devices and technologies, software in the 1960s-70s underwent significant changes. In particular, the paradigm of program execution independence from the platform, key ideas of graphical interfaces, and keyboard-friendly shell applications were developed. The first examples of software with virtual reality elements were implemented, such as object control in virtual space, computer games, and AI assistants that simulated the behavior of specialists.

The Universal Programming Language C. Developed by Dennis Ritchie at AT&T Bell Labs in 1972, one of the key advantages of C was the ability to write programs independent of specific hardware platforms. This was achieved through a higher level of abstraction compared to assembly languages. This abstraction allowed programmers to use high-level language constructs to write universal code that was then compiled into machine code executable on different devices.

Compiling commands into machine-independent code made C the foundation for creating many other programming languages and operating systems, including those that support graphical interfaces. This provided high flexibility and efficiency in software development, making C one of history's most significant programming languages.

Shell Command Terminal. In the 1970s, the UNIX operating system was developed, revolutionizing programming and data management approaches. UNIX introduced the concept of Shell — an interactive command interpreter that evolved from a simple command execution tool into a powerful programming instrument. Shell allowed users and developers to create complex scripts and automate tasks, significantly speeding up the development and management process. These features became so fundamental that their elements are preserved in modern operating systems, even those equipped with graphical interfaces.

Graphical Interfaces.  One of the first examples of a graphical user interface (GUI) was the Xerox Alto, introduced by the company in the early 1970s. Using a mouse pointer, the computer demonstrated a new way of interacting through graphical images, windows, and icons. This revolutionary step significantly simplified computer usage and made them more understandable for a wide audience.

In the late 1960s, Terry Winograd at MIT created SHRDLU, which allowed manipulating a virtual world using natural English language expressions and moving objects such as blocks and cones in a simulated physical environment. These studies were further developed in the “Put-That-There” system (MIT, 1979), which allowed the control of virtual objects on the screen using gestures and voice commands without needing a keyboard.

Behavior-Imitating Programs: ELIZA, developed by Joseph Weizenbaum in the mid-1960s, became one of the first programs to allow conversation with a computer in English. The program imitated the active listening technique of a psychotherapist, responding to user questions based on keywords identified in the user’s remarks. Early AI research showed that machines could perform specific tasks on human commands, engage in dialogue, adapt to human requests, and even anticipate user needs.

Computer Games. Text-based quests and adventure games defined the development of many mechanics of dialogue interfaces. Games such as “Star Trek,” “Adventure,” and “Hunt the Wumpus” not only entertained but also contributed to developing logical thinking and imagination skills. Unlike early graphical games like Spacewar! (1962), which required specialized computers and complex equipment, text-based games could run on simple personal computers. These games relied on textual descriptions of the world and scenarios, allowing players to interact with the game through text commands.

Due to limited technological capabilities, text-based games laid the foundation for the interactive fiction genre and demonstrated the possibilities of interacting with a computer through natural language dialogue. They showed that even without complex graphics, creating an engaging and intellectually stimulating gaming experience was possible, which became an important step in developing computer entertainment and interfaces.

Conclusion

The new ideas and engineering solutions of the 1960s and 1970s became the foundation for ergonomic principles in interface design. Human factors, ease of use, and visual clarity began to be considered. As a result of these practices, it became easier for people to interact with computers.

The emergence of game consoles such as the Atari 2600 greatly impacted popular culture, turning video games from a niche hobby for enthusiasts into a mass pastime. Games like Pac-Man became cultural phenomena. Home gaming systems allowed millions to get acquainted with digital technologies for the first time and understand their potential.

These decades laid down the principles defining how we interact with the digital world today. The first personal computers, operating systems, internet protocols, interface devices, and touch screens were created. Even video conferencing, which was met with skepticism then, is now a common means of business and personal communication. As a result, digital technologies have become a convenient tool and a catalyst for global changes in society, altering how we work, communicate, and entertain ourselves.

The foundations of human-machine interaction were laid in the era of the first tools and mechanisms. In the pre-computer era, human-machine interfaces were limited to mechanical and analog devices, but even then they reflected the principle of necessary diversity formulated later by cyberneticians.

This principle states that the richer the range of control of an information system and the more precisely the control actions are coordinated, the greater the efficiency and adaptability of the system. In the context of interface development, this emphasizes the critical importance of creating solutions intuitive to users and capable of performing complex tasks.

The term "interface" in this historical account goes beyond simple human-computer interaction to encompass the entire arsenal of tools, mechanisms, and techniques that have enabled humans to control machines and processes. Interfaces built bridges between the physical and mechanical worlds, both by means of input tools - levers, buttons, and keys - and by means of information displays - pointers, scales, and audible signals.

Since ancient times, interfaces have been divided into professional and consumer interfaces, each reflecting different user knowledge and skills requirements. From command lines requiring specialized expertise to intuitive search engines and devices accessible to everyone, the evolution of interfaces demonstrates the constant desire to simplify human interaction with technology.

With this article, we begin a journey back in time, exploring how primary mechanical devices and early technological solutions formed the basis for today's interaction systems. We will start with the pre-computer era when keys and locks carried a physical function and symbolized ideas of security and access. Every tool and mechanism had an immediate, tangible meaning. Immersing ourselves in this era will help us understand how the fundamental principles of human-machine interaction remain relevant today despite dramatic changes in technology and design approaches.

Simple mechanisms for interacting with machines in the pre-computer era

Multiplication of physical force. Many early human-machine interfaces were based on simple mechanisms such as levers, blocks, and gears that played a crucial role in the development of civilization. These elements not only multiplied human physical strength but also opened up new possibilities for construction, transportation, and military technology. Using these mechanisms in ancient lifting devices and water lifting mechanisms does not simply demonstrate early principles of force transmission and amplification; it laid the foundation for the subsequent development of engineering and mechanics.

Information Display. In the pre-computer era, information display mechanisms were closely associated with using various indicators such as scales and hands. Compasses, astrolabes, sundials, water clocks, and mechanical clocks used these elements to provide visually perceptible information about physical quantities such as time, distance, and speed. The scales and hands on these devices allowed people to visually evaluate and interpret the measured parameters, enabling effective navigation, planning, and coordination of activities. Thus, these elements acted as critical components of early information interfaces

Mechanical computers and musical instruments demonstrate the principle that form follows function. These devices are designed so that their structure and control mechanisms mimic the principles of what they are intended to measure or control. For example, the division of a score in an abacus or the arrangement of keys on musical instruments reflects the digitization of numbers or octaves in music, emphasizing how the device is adapted to control a particular order or structure. That is, the physical structure of devices is directly related to their functional purpose. Visual and tactile design serves to better understand and interact with measurable and controllable phenomena.

Automation: forward and feedback. Transportation and industrial devices such as steam locomotives and textile machines demonstrate the development of human-machine interfaces in the context of forward and feedback. Controlling these machines required operators to understand control mechanisms and interpret feedback signals, enabling real-time adaptation and adjustment. This interaction not only contributed to improved operational efficiency and safety but also laid the groundwork for the development of the control systems and interfaces of the future.

Predecessors of text-based interfaces. The typewriter, first invented in the 19th century, was one of the key technological breakthroughs in the history of human-machine interaction. It provided the ability to convert thoughts into printed text rapidly and standardized, greatly speeding up and simplifying the process of creating documents, letters, and literary works. The typewriter interface, while requiring some skill to use effectively, was intuitive and allowed users to see their actions' results on paper immediately. This immediate feedback between action (pressing a key) and reaction (the appearance of a symbol on paper) is a fundamental principle of many modern interfaces.

Security interfaces. The key-lock mechanism can be seen as one of the oldest analogs of security interfaces for controlling access to something. This system is a primitive form of user authentication, where the key serves as a physical "password" that authorizes or denies interaction with the lock. The concept of privacy and security remains relevant in today's technological interfaces. We can even use a hardware security key physically attached to the device.

Coding and signaling systems. Ancient signaling systems, such as transmitting messages through signal lights or semaphores, were used to communicate over long distances long before the invention of electrical devices. These methods relied on visual perception and required the sender and receiver to know a predetermined code (e.g., a certain number of lights or the hand position of a semaphore could signify a particular message).

The telegraph revolutionized long-distance communication by enabling instantaneous transmission of messages over long distances using electrical signals. The Morse system, adopted to encode messages, was one of the first standardized interfaces for human-machine interaction, where short and long signals (dots and dashes) were combined to represent letters and numbers. The telegraph demonstrates an important step in the evolution of interfaces, where a mechanical action (pressing a key) is converted into an electrical signal, which is then interpreted as a text message.

Punched cards were one of the earliest ways to input data and programs into computing systems. Punched cards were rectangular cards made of heavy paper or cardboard. Data was entered into them manually using a special device called a punch card. The operator placed a blank card in the puncher and pressed the keys arranged like a conventional typewriter. Each key press controlled a die that punched a hole at the appropriate position on the card. The positions of the holes were strictly defined, and various information, such as numbers, letters, commands, etc., were encoded.

Typically, a punch card had 80 or 96 columns (columns), in each of which one of several holes representing numbers or letters could be punched. Combinations of holes in different columns made up the encoded data. In modern linters we can find similar limitations on the number of columns of program code.

Special devices — punch card readers or simply readers — were used to read data from punch cards. They literally "read" the location of the holes by means of metal pins passing through the holes, translating the physical holes into electrical signals that could be processed by computers.

Although primitive by today's standards, the punch card interface marked an important step in developing data entry methods. Punched cards made automating the data entry process possible, replacing manual input entirely. They were widely used in the 1960s and 1970s to program early computers.

Analog computing machines

Analog computers were early mechanical and electronic systems used to model and solve complex mathematical problems by physically manipulating quantities. These computing devices used continuous physical processes, such as gears' rotation, fluid flow, or voltage changes, to represent and process numerical values.

The differential analyzer, developed by Vannevar Bush in the 1930s at the Massachusetts Institute of Technology, was one of the best-known mechanical analog computers. This device consisted of a complex system of integrated components, including torsion drives, differential mechanisms, and automatic recording devices. Users could set the initial values of variables and the relationships between them using mechanical levers, buttons, and scales. The computer calculated the results, displaying them on rotating graphs and charts. This interface allowed researchers to visualize and analyze complex mathematical relationships using the mechanical analogy of motion and relationships between components.  

Analog computing machines, popular in the 1940s and 1950s, used electronic components such as operational amplifiers to simulate differential equations and other mathematical concepts. Users set initial conditions and parameters using potentiometers, switches, and other manual controls. Computational results were displayed on oscilloscopes and arrow indicators. These electronic computers provided greater accuracy and speed of computation than their mechanical counterparts, but their interfaces were still based on physical representations of quantities and required a thorough understanding of mathematical principles for effective use.

First approaches to the realization of human-machine interaction in the 1940s-50s

The ENIAC was the first general-purpose electronic digital computer that could be reprogrammed for various tasks. The computer was implemented on electronic tubes and was essentially an electronic version of a large mechanical digital analyzer. The architecture of the computer appeared during World War II as a solution for computational problems associated with the adoption of new types of weapons, taking into account their firing quality, and later - as a calculation machine for the tasks of modeling a thermonuclear explosion. In fact, the machine was intended for solving mathematical differential equations. Intermediate results were displayed on punched cards and after recomputation were again entered into the machine.

The mid-twentieth century ushered in an era of research in human-machine interaction that opened the door to developing systems that could mimic the human mind and associative thinking. One of the early breakthroughs was Vannevar Bush's article "As We May Think," published in 1945, which presented the concept of a hypertext system. This system, called Memex, was envisioned as a device for storing vast amounts of information and retrieving it conveniently, which could radically improve knowledge management.

Bush proposed the use of electromechanical mechanisms to control reading and writing on microfilm, which allowed the user not just to save documents and notes but also to create complex networks of associative links between them. This idea, anticipating modern hypertext and Internet technologies, emphasized the importance of convenient access to and organization of information.

The generation of sounds close to speech began to be solved in the 30s-40s. The first major breakthrough was the Voder. The machine synthesized human speech by mimicking the effects of the human speech pathway. Using a handle on the wrist a system of keys, the operator could simulate consonant and vowel sounds, controlling the pitch of the sound with a foot pedal. It was a difficult machine to operate, but after several months of practice, a trained operator could produce recognizable speech.

A notable alternative to controlling devices via keys and buttons was voice control, which began to take hold in the 1950s, when the first attempts at voice control of computers were made. One of the pioneers was Audrey, a system developed in 1952 by Bell Labs engineers.

The Audrey system could determine the digits from 0 to 9 with an accuracy of 98%. This was a real breakthrough for that time - the first transistor computers appeared only in the second half of the fifties.

Conclusion

By looking at the evolution of human-machine interfaces in the pre-computer era, we can trace a continuous drive toward ever more natural, intuitive, and powerful means of interaction. Early interfaces, such as levers, blocks, and gears, provided humans with the ability to amplify and direct their physical abilities with the help of machines. As technology evolved, interfaces began to extend beyond mechanical systems to include methods of information exchange, authentication, data display, and even the rudiments of programming.

We are now seeing ever closer integration of man and machine, where interfaces are becoming virtually invisible, allowing natural interaction through voice commands, gestures, and even neural signals. Such aspirations are going in the direction described in science fiction in the 40s and 60s.

The development of interfaces not only expands the possibilities of interaction but also has a profound impact on human nature itself. Interfaces shape the way we perceive and interact with the world around us, changing our habits, behaviors, and cognitive abilities. The design of new interfaces must take into account both technical and social, ethical and humanistic aspects to ensure the harmonious coexistence of man and machine in the endlessly evolving world of technology.

Bubble is a platform for building web applications using visual programming. The project was started in 2012 and is one of the pioneers in the NoCode field. Over two decades, Bubble has accumulated considerable experience, and most of the solutions offered by the company have already passed the test of time. In the NoCode community, the platform is one of the most mentioned and used tools.

Below, we will look at how Bubble works, the differences between this platform and similar solutions, and examples of successful commercial projects created with Bubble.

How to make a web application on Bubble

The critical feature of Bubble is its powerful and intuitive visual editor. It allows you to create complex web applications with adaptive interfaces and visually build workflows with interactive elements.

Bubble's basic principle is visual programming. While it has many ready-made templates, the tool has a flexible design. The interface of the created application can be customized down to minor details, making it unique and professional-looking.

The application development process in Bubble typically consists of four steps:

1. Visual Programming. Using a visual editor, the user creates an interface by placing controls- buttons, text fields, and images- on the canvas.
2. Logic and database customization. After creating the interface, the user customizes the application logic - what events occur when clicking on certain parts of the layout, what data is sent, and where.
3. Automatic code generation. Bubble automatically generates the necessary code to implement the interface and logic created in the previous steps: visual CSS and HTML blocks and JavaScript code for active elements.
4. Testing and deployment. In the final stage, the application can be tested on the platform and deployed to the web. Bubble provides both hosting and server infrastructure management. It should be noted that application deployment is not included in the free tariff - to publish the project on the Internet, you will need to pay a subscription.

The development process in Bubble does not require programming skills at any of the listed stages. As a result, the platform allows you to quickly create working web applications and fully control their appearance and functionality.

How Bubble differs from other NoCode creation services

Bubble differs from similar web application development platforms in several key ways. Most of them can be found in one project or another, but in the case of Bubble, the time of the platform's existence and the degree of development of each approach are essential.

Built-in features. Bubble has many built-in features that provide basic functionality for most applications:

• email validation,
• file storage,
• progress bars,
• adding captcha,
• localized translations,
• side menus,
• payment processors.

These solutions are designed as standardized blocks with familiar working mechanics. These features simplify the development process, preventing users from worrying about the underlying application infrastructure.

Plugins significantly extend Bubble's capabilities by providing more specific elements, actions, and APIs than built-in functions. These are often more extensive solutions with documentation on the platform. Here are a few favored groups of plugins and individual examples:

• Data import/export: creation and loading of CSV files and other data files, including work with external databases;
• utilities: API Connector allows you to connect to any API on the Internet and exchange data with your application. This gives you more flexibility than using specific plugins to link to specific services;
• integration with third-party services: Airtable, Facebook, Figma, YouTube, Google Maps, Twitter, LinkedIn, Pinterest and Google (full list);
• payment systems: Stripe and PayPal plugins simplify the process of accepting payments in the application;
• barcode and QR code generation: plugins for creating or reading unique identifiers for users, products, or other data.

Since plugins are a more specific solution, they are installed and configured separately in the Plugins tab. Some plugins are created and described by the Bubble team, while others are developed by the Bubble community and companies that use the product.

Most plugins are free, but some require payment to install or may require adding a key to authenticate the connection between your account on the service and the plugin in Bubble. In the latter case, the tariff is determined at the service level.

Built-in database. Unlike many other NoCode platforms, Bubble offers extensive database capabilities. Users can create, modify, and manage complex data structures without writing SQL queries. Database design is as visual as the interface.

Users can create their own data types, essentially tables in traditional databases. These data types can contain various fields, including text, numbers, dates, files, and even relationships with other data types. Complex structures and associations can be modeled as in classical database management systems, such as one-to-many or many-to-many. The data documentation section describes the basic principles of working with data.

Templates. The platform has about 1,500 ready-made templates and examples - easy to create for a marketplace, delivery service, or online school. Templates include pre-designed screens for specific purposes and workflows that support data flow and logic within the application. Templates, just like plugins, are subject to the mechanics of the marketplace: they can be free or paid and are usually supported by specific developer contributors.

Collaboration. Bubble has a convenient way of working with versions, similar to the Git version control that has proven itself in classic development. You can split the story into branches and then merge the work of individual contributors into a single view.

Community. Compared to alternative platforms, Bubble has one of the largest developer communities with an open forum. This allows you to solve problems and find ready-made solutions quickly. There is also an extensive database of tutorial resources, which makes development on the platform accessible even for beginners.

Flexibility and Scalability. Bubble allows you to create scaled applications with your growing business needs. The platform is suitable for small projects and complex systems, which is reflected in the project's fee schedule.

Examples of successful commercial projects based on Bubble

The Bubble platform has enabled the creation of many successful commercial projects:

• Norebase makes it easy to register trademarks in Africa.
• Moone solves the problem of tight work schedules by offering tools for personal and team scheduling.
• Swapstack was one of the first successful projects that simplified the monetization of newsletters by attracting advertisers and sponsors. The project was taken over by the NoCode-based newsletter platform Beehiiv.
• Goodgigs helps green companies find employees from their community. In addition to the job search, Goodgigs offers a separate space for socializing and weekly updates on job openings.
• AwareHealth is a healthcare platform for companies focused on personalized care for employees.
• LifeLegacy simplifies financial preparation for various life events, making writing wills, financial powers of attorney, and other documents easier.

For examples of other companies, check 30 Bubble-based succes stories article.

Conclusion

Bubble was founded in 2012 by Joshua Haas and Emmanuel Strisch. The main idea behind the project was to enable users without technical skills to create potent web applications. At the beginning of its existence, Bubble focused on simplifying the web development process by making it accessible to entrepreneurs, marketers, designers, and non-programmers.  Bubble is currently the leader in functionality among NoCode web application builders.

However, it's worth noting that Bubble's confident leadership has a downside:

• Bubble's paid plans are expensive compared to its competitors. For example, Webflow's basic tariff costs $14 versus Bubble's $29, but even with the free tariff you can publish a couple of pages for a test;
• A well-developed ecosystem creates the risk of vendor dependency, but Bubble apps can't be ported to other platforms;
• Bubble focuses on web application specifics and is inefficient regarding atypical backend logic or optimizing complex database queries. Performance and scaling issues can arise under heavy loads.

When considering the platform's limitations, Bubble is a convenient solution for small teams and startups to build web applications quickly.

Game development is a collaborative and intensive endeavor, bringing together a diverse mix of professionals - from programmers and designers to artists. Each game blends engaging stories, rich visuals, and intricate software solutions. Recognizing the challenges in coding and development efficiency, GameDev pioneers have increasingly turned to NoCode tools. These tools empower scene and level designers to build game worlds with simple drag-and-drop interfaces, eliminating the need for coding. This shift allows developers to focus more on creative aspects like design and gameplay.

This article explores the significant advantages of NoCode tools in game development. They simplify the entry into game development, making it accessible for beginners and non-programmers. This approach fosters creativity among designers and artists and accelerates growth and content creation. By leveraging pre-built templates and machine learning, developers can concentrate on the creative side of games, such as storylines and gameplay dynamics.

NoCode tools have been game changers in the industry, democratizing game development and enabling a wider range of creators to bring their unique visions to life. From fostering innovation in game design to streamlining the development process, NoCode tools mark a significant evolution in the gaming industry that began taking shape in the 1990s.

First examples of creating parts of games with NoCode-mechanics

The GameDev industry has been actively utilizing the power of NoCode for decades. Even early id Software games such as Doom (1993) and Quake (1996) provided easy to learn tools for players to create their maps, levels, and modifications. These were some of the first examples of players being able to participate in the creation of game content actively. The appearance of such editors influenced the future development of modding in Half-Life and The Elder Scrolls. Today, the process of creating game levels has taken a new turn, and in games, you can find automated generation of quests and locations or real-time age of worlds: No Man's Sky, Terraria, Elite Dangerous.

The idea of democratizing content development was further embodied in games like Minecraft (2011), LittleBigPlanet (2008), and Super Mario Maker (2015). For example, in the latter, a year after its release, players worldwide had created more than seven million levels and played more than 600 million hours. Built-in level and object builders allow you to create from scratch without special skills, not only unique worlds and new game mechanics. With Steam Workshop you can share the created content with other players.

Modern GameDev platforms also allow developers to create games without having to write code, making game development more accessible and flexible.

GameDev platforms using NoCode mechanics

The two most popular and largest competing platforms for game development are considered to be Unity and Unreal Engine. These are mature software solutions with plugin stores developed by third-party GameDev studios - including plugins with support for NoCode mechanics.

Unity is a game development platform based on the C# programming language, in which visual programming concepts are implemented both at the root level in Unity Visual Scripting and in the form of special Playmaker and NodeCanvas plugins.

• Unity Visual Scripting is a visual graphics engine and programming system that grew from the Bolt plugin and evolved into an embedded Unity solution by 2021. The system allows developers to create game logic and interaction without writing C# code. The game designer manages game events, states, and character behavior by creating an event graph.
• Playmaker is a popular paid plugin developed by Hutong Games. Playmaker has a huge community of users and many add-ons.
• NodeCanvas - Paradox Notion's plugin provides a flexible graphical interface with different types of nodes, including decision trees, states, and action sequences

Unreal Engine, unlike Unity, uses C++ as its primary programming language but also provides several tools for visual programming. The platform is known for its graphics engine, which can be used to create photorealistic graphic scenes.

Here are some examples of specific subsystems and plugins in Unreal Engine that utilize visual programming:

• Blueprints is a built-in Unreal Engine system that allows you to create game logic and interaction using visual graphs. Developers can create Blueprint classes to define events, variables, functions, and many other elements of game logic using blocks and nodes. It is actively used in both indie projects and AAA games.
• Niagara is a visual programming system for creating effects related to particles: fire, smoke, gas, and water. With its help, developers can create complex and colorful visual effects.
• Behavior Trees and Blackboard is a two-asset system for describing the behavior of game characters: with Behavior Trees, developers can define the behavior of NPCs and monsters in the game through visual decision trees, while Blackboard allows you to define variables and data for use in Behavior Trees. There is a Behavior Tree Quick Start Guide available for review.
• Sequencer is a visual storytelling tool that creates cinematic scenes, complex camera movements, and animations.

In addition to the two popular giants, there are smaller solutions that typically have fewer features but a more gentle learning curve:

• Construct 3 is specialized for creating 2D games and provides an intuitive interface for developing such games. This makes it an excellent choice for quick 2D projects. Examples of games include Ultimate Chicken Horse and Forgotten Worlds.
• Core Games is a web platform for creating and publishing multiplayer online games, which is its main advantage. Examples of games: Forgotten Cisterns, Tumbleweed Typo Hunters
• Roblox Studio - Emphasizes social interactions in games, where players can interact with others in created worlds and games. This platform also provides NoCode tools for game creation and an extensive library of assets. Examples of games: Rainbow Friends, Blox Fruits,
• GameMaker supports game development for various platforms, including iOS, Android, PC, and consoles. The platform became a versatile tool for developers looking to reach the largest possible audience. Game examples: Undertale, Forager
• Tynker is an educational platform for creating games and animations that provide blocks and a visual interface for creating game logic. The project is focused on teaching children how to program and create their games.
• Godot is a free, open-source game creation engine that allows you to both write code and has tools for NoCode development. An example of a popular game is Doki Doki Literature Club!

Thanks to the availability of such alternatives, the doors to the GameDev industry are now more comprehensive than ever. Development teams can choose the right platform according to their experience, budget, and target audience.

NoCode platforms for creating games for iOS and Android

There are also specialized NoCode tools for creating games for iOS and Android:

• Buildbox specializes in creating games with an intuitive design and provides templates and ready-made components to simplify the game creation process.
• GameSalad is focused on creating games with more complex logic and controls. It provides the ability to develop scenarios and game logic with a higher level of control.
• GDevelop is a free and open-source NoCode tool. It provides a flexible visual interface and supports exporting games to various platforms. GDevelop is also supported by an active developer community and has extensive documentation. The game Epic Battle Fantasy 4 was created using GDevelop.

Most such tools have additional functionality for publishing games to the App Store and Google Play.

Auxiliary NoCode tools for GameDev

Autodesk Maya Bifrost is a system for creating complex simulations, animations, and visual effects in Autodesk Maya (simulation of liquids, smoke, fire, and fabrics). One of its parts is the Bifrost Graph, which provides a visual interface for creating complex visual effects and animations without writing code. Developers can use the GUI to create complex animation logic and impact, which can help create impressive game animations.

Blender is a free and open-source 3D graphics package that also provides a tool called Animation Nodes, which allows you to create complex animations and scene logic using visual programming. This can significantly simplify creating animations and interactive scenes without writing code.

Use Spine to create sprite and character animations. The tool provides a visual interface for creating complex animations that can be easily integrated into games.

Use Bfxr to generate sound effects. The tool provides an interface for creating and customizing sound effects that can be easily exported and used in games.

These tools allow you to create a variety of game resources and content without the need for programming and can be useful support tools in the game development process.

Opportunities and limitations of NoCode technologies in GameDev

In the previous sections, we looked at how NoCode technologies are revolutionizing approaches to game development, opening new horizons for designers, artists, and novice developers. These tools significantly speed up the creation process and allow you to realize creative ideas without deep programming knowledge.

Yet, like all technologies, NoCode has limitations that need to be weighed when selecting a game development approach. We'll delve into what NoCode tools can effectively achieve in GameDev and identify scenarios where traditional programming might be necessary.

Opportunities with NoCode:

1. Prototyping and Design. NoCode tools excel in rapid prototyping and designing game levels. They enable designers and non-programmers to bring their ideas to life without extensive technical expertise.
2. Developing game logic and mechanics. Visual programming capabilities of NoCode tools simplify the creation of basic game logic and mechanics. This includes designing character movements, interactions, elementary AI patterns, and straightforward game scenarios.
3. Standard solutions integration. NoCode facilitates the seamless integration of essential game resources like graphics, animations, sounds, and music, bypassing complex coding.
4. Leveraging Large Language Models. These models can generate narrative content - from dialogues to storylines, thereby expediting the game development process.

Limitations of NoCode:

1. Complex logic and deep customization. When developing advanced AI, intricate physics, or deep customization, NoCode tools might fall short, necessitating manual coding for such sophisticated functionalities.
2. Optimization and performance. For large-scale, resource-heavy games, achieving optimal performance often requires code-level optimization, which might exceed the scope of NoCode platforms.
3. Unique innovation and customized design. Although NoCode tools provide a broad spectrum of functionalities, they might restrict creative freedom in developing highly unique and innovative game elements that demand a tailored approach.
4. Integration of advanced technologies. Incorporating complex technologies, especially in advanced AR/VR areas, typically requires specialized programming expertise, a domain where NoCode tools may not fully suffice.

In general, it can be said that the GameDev industry is actively evolving, integrating cutting-edge technologies with classic approaches to create immersive, interactive experiences. And in this process, NoCode's concepts and tools are increasingly important.

The Internet of Things (IoT) is a computing network of physical devices equipped with technologies to communicate with each other or the external environment. The modules of this network can be small sensors and actuators, as well as composite systems such as smart homes. According to Statista, the number of connected IoT devices will reach 12.3 billion in 2022, which continues to grow.

A key feature of IoT is data collection and exchange between the devices themselves and some centralized services. Somebody can use this data to monitor, control, and optimize everyday life and business aspects. For example, smart sensors can collect information about temperature, humidity, lighting, and other parameters in real time and then transmit this information to a central system that analyzes the data and makes appropriate decisions.

IoT has a wide range of applications:

• Smart Spaces. IoT devices can control lighting, temperature, security, and other aspects of the home. In the smart city concept, IoT manages transportation infrastructure, improves public safety, and optimizes energy consumption and urban life.
• Healthcare. In the medical field, IoT is used to monitor patients' vital signs using specialized equipment and wearable trackers, manage medical devices, and develop new methods of diagnosis and treatment.
• Industry and agriculture. IoT helps automate manufacturing processes, improve efficiency and safety, predict potential breakdowns, and reduce costs. IoT helps monitor soil, plants, and livestock in agriculture to increase yields and reduce costs.
• Transportation and logistics. IoT devices track the location and condition of vehicles, optimize logistics and improve road safety, and move goods and postal items indoors.

Let's look at a few examples of projects that utilize IoT technologies.

Examples of IoT projects

Below we have provided some inspiring examples of projects in IoT of varying degrees of extravagance.

The smart home is the most famous example of a low entry threshold. Sensors collect data and transmit it to some control panel to provide efficient house control. Goals can range from saving energy to automating routine tasks such as adjusting the heating based on temperature. A smart home system can display the family calendar, to-do list, weather conditions, and even transportation status on the panel – handy for a sneak peek before going out in town. You can use mini-PCs or miniature boards like Raspberry Pi, just a phone or tablet with a suitable dashboard program to control such a system.

Sound Detector. If you have pets, such as dogs, and you are concerned that they are sad when you leave the house, you can create a system that will reassure the animal by triggering a specific recording when the dog starts whining, howling, or moving around the room too quickly.

An office or houseplant care system. This project involves a system that monitors office conditions, including lighting, temperature, and humidity, and controls the watering and lighting of plants to create a comfortable and healthy work environment. Soil moisture sensors will detect when plants need to be watered and send an alert or start the auto irrigation system.

AI-powered smart pet door. One of the popular combinations in the IoT world is the combination of artificial intelligence, some sensors to gather information, and mechanisms to control the situation. A famous example is the automatic opening of a door when the owner's cat appears in front of a camera, filming the space in front of the door.

These are just a few examples. You can explore resources such as Instructables and other sites dedicated to IoT and DIY technologies for inspiration for additional projects.

IoT devices that can be interacted with using NoCode systems

You already have IoT devices, but you may not know it: for example, one of the most common IoT devices is the smartphone. They are usually divided into consumer and industrial:

• Consumer IoT devices are smartphones, smart watches, trackers, wearable devices, intelligent assistant speakers, and home appliances. Such are designed to be used in everyday life, making it more convenient and comfortable.
• Industrial IoT devices are focused on one business use or another: manufacturing, medicine, agriculture, automotive and logistics, lighting, and parking systems.

In the Internet of Things (IoT), several vital standards play an essential role in ensuring device interoperability and security:

• Zigbee is a standard for low-energy mesh networking (devices form a peer-to-peer network) used by companies like IKEA and Bosch.
• Z-Wave is a Zigbee-like standard that has gained much popularity in the US.
• Wi-Fi + cloud services – although Wi-Fi is not specific to IoT, it is widely used in this area.

Popular IoT Devices:

• Smart outlets control appliances based on scenarios, for example, depending on the geo-position of the phone.
• Home appliances: robot vacuum cleaners, refrigerators with inventory monitoring, access control systems, humidifiers and air purifiers, and coffee machines.
• Temperature, humidity, carbon dioxide sensors, and thermostats for heating savings and climate control.
• Light sensor for street lighting after sunset.
• Motion sensors to control lighting or security cameras.
• Door closing sensors, smoke detectors, and security systems for added safety.

Also, consider using conventional smartphones to control the system with voice commands or to trigger commands via an external trigger message.

The potential of using the NoCode approach in the IoT

NoCode and LowCode technologies provide an opportunity to accelerate the development and deployment of IoT projects, reducing dependence on professional developers and programmers:

1. Building applications and systems faster and easier. NoCode and LowCode platforms allow even people without deep technical experience to create IoT applications and systems. They provide graphical interfaces and tools to build applications by dragging, dropping, and customizing components. Instead of writing code from scratch, developers can use off-the-shelf components and templates, which reduces development and maintenance costs.
2. Integration with external systems. NoCode and LowCode platforms typically provide tools for integration with other systems and services. This makes linking IoT devices with cloud services, databases, and management systems easy.
3. Enhanced data management. With these technologies, you can create dashboards and interfaces to monitor and manage data collected from IoT devices.
4. Prototype and MVP development. NoCode and LowCode enable rapid prototyping and minimum viable products (MVPs) to quickly assess the potential of an IoT project idea and attract investors or customers. With rapidly evolving AI solutions, LowCode and NoCode look like excellent foundations for software solutions, while IoT looks like a foundation for hardware solutions.
5. Risk mitigation. With visual development tools and pre-testing, you can mitigate the risks associated with developing IoT solutions and make quick project changes when you reuse IoT modules for new tasks.

However, it is worth noting that NoCode and LowCode technologies can have their limitations, especially in the case of complex and specific IoT projects. In such cases, deeper programming and customization may be required, and NoCode/LowCode platforms may need more flexibility.

Overview of popular NoCode platforms for IoT

NoCode platforms provide the user with the ability to create automation chains. The platforms offer templates for collecting and processing data flow, monitoring events, and sending notifications at specific triggers. However, the systems target different user groups and are designed for different usage niches.

IFTTT (If This Then That) is a web service that allows users to create chains of conditional statements called "applets". These applets automatically execute tasks based on specified conditions. The IFTTT library has many out-of-the-box solutions for integrating IoT devices (smart lights, thermostats, security cameras) and web services (social networks, email services). For example, you can set up an applet in your home to automatically turn on intelligent bulbs when the sun goes down. IFTTT is aimed at a broad audience and allows you to create chains without programming knowledge.

Node-RED is a Low-Code visual programming tool based on Node.js. It allows users to connect devices, APIs, and online services using streaming. Node-RED is widely used in industrial and hobby IoT projects to create complex workflows. It enables various devices and services to integrate, providing more profound control and customization capabilities than IFTTT. Node-RED is aimed at those with programming experience or willingness to learn. It is a more professional-grade platform that allows you to create more complex and customizable automation. You can read more about Node-RED in the article Node-RED for App Development.

ThingsBoard is a specialized Open-source IoT platform for data collection, processing, visualization, and device management for the Internet of Things. Unlike IFTTT and Node-RED, this platform is designed for IoT tasks only. ThingsBoard supports various IoT protocols, including MQTT, CoAP, and HTTP, and offers built-in data analysis and visualization tools. This platform targets developers and enterprises looking for a scalable and extensible solution to manage and monitor a network of IoT devices.

Waylay is the Enterprise counterpart to the open source ThingsBoard solution with support for the LowCode paradigm and the ability to orchestrate workflow at the enterprise level. The platform also implements an abstraction layer on top of traditional IoT platforms, enabling the interconnection of cloud-based IoT solutions.

IoT-NoCode companion platforms

Developing and managing Internet of Things (IoT) systems is simplified through synergies between device manufacturers and NoCode/LowCode platforms. This approach makes it much easier for users with deep technical programming knowledge to create and configure intelligent systems, and often, NoCode platforms only try to cover some possible IoT devices but focus on a specific niche. This is how combinations of companion platforms emerge that are most convenient to use together.

Tibbo Project System and AppBlocks. Tibbo offers the Tibbo Project System (TPS), which consists of modular Tibbits. These Tibbits come in different types, including I/O modules, connectors, and hybrid devices that provide a wide range of functionality. They are designed to create personalized IoT solutions by allowing the integration of various components into a single system. Tibbo provides out-of-the-box applications and programming capabilities to customize these modules, simplifying the creation of complete IoT solutions. The AppBlocks companion platform is a NoCode platform for developing IoT applications running on TPS devices. It allows users to build IoT and automation solutions in the browser without writing code. AppBlocks uses a visual designer to define application logic and supports debugging directly in the browser.

Philips Hue and IFTTT. Philips Hue offers a wide range of intelligent lighting solutions. IFTTT, a platform that allows you to create simple automation (e.g., turn lights on when the sun goes down) without the need for programming, can be used to extend the capabilities of Philips Hue devices. IFTTT supports other devices as well.

Arduino and Node-RED. Arduino is widely recognized in DIY and education for its easy-to-use microcontrollers and IoT modules. Node-RED is a visual programming tool that makes connecting devices, APIs, and online services easy. Together, they provide powerful tools for creating custom IoT solutions. This combination proves to be incredibly flexible if you have the time to understand the features of both platforms – just as many different add-ons have been created for Arduino, Node-RED has many ready-made libraries to work with other systems, including Arduino.

Conclusion

Integrating IoT and NoCode democratizes the creation of intelligent systems and devices, allowing the system to be built from modules at both the software (NoCode) and hardware (IoT devices) levels. Users can experiment and realize ideas that would have been inaccessible to them due to technical barriers.

NoCode and LowCode platforms simplify the development of IoT solutions by enabling faster prototyping, integration of different devices and services, and data management. This opens up opportunities for technical experts and users who need programming experience.

Due to the specific need for electronics knowledge, the implementation of complex projects still requires the involvement of professional developers and engineers. However, as AI and machine learning technologies integrated into such platforms evolve, the barriers to creating IoT solutions will decrease.

The rapid development of NoCode tools means community members do not have time to master each tool. Documentation and accompanying materials need to catch up. As a result, the world of NoCode technology elicits mixed reactions, ranging from inspiration to frustration, and the environment can seem overly chaotic and unstable.

In such an environment, building connections within the community is more effective than delving into individual tools. NoCode principles aim to make technology accessible to a wide range of people. However, the primary audience of NoCode projects is interested developers who are the first to master new tools and know the general techniques of working with them better than other users. Unlike traditional programming, in the world of NoCode, the community plays a crucial role in solving technical issues.

Interaction with the community not only helps to develop specific projects but also shapes the future of technology. Here, you can find fresh ideas and requests to create new tools, establish connections with like-minded people, and expand your professional network of acquaintances: find business partners or a team to implement projects.

At the same time, the NoCode community is diverse and includes various subgroups and organizations. Some focus on training and providing educational content for beginners, while others bring together experienced developers to collaborate on projects.

There are communities with open membership and those where you must apply. Some groups offer members access to valuable tools and credits on popular NoCode platforms. Below, we will take a closer look at a few examples of such communities, categorizing them into subgroups:

• informational communities are built around a group of people interested in a variety of information related to NoCode,
• product communities bring together users of a particular product or multiple products,
• educational communities form educational products about NoCode technologies for other members of the group,
• HR platforms where NoCode developers and projects find each other.

Informational NoCode Communities

NoCode communities, in the broader sense, do not have a single platform for exchanging ideas. Discussions are often organized in Twitter and Slack channels, and communities rarely have a single discussion channel. The results of news summarization can be received through newsletters.

NoCodeDevs is an open community with more than 10 thousand members. It sends out weekly news about NoCode tools, it also has some tutorials, merch, a community forum and a blog. The platform also offers $99 educational courses, each focusing on solving a different case study from scratch to product creation.

NoCodeOps is a community of over a thousand NoCode developers offering workshops and joint sessions. To join the community, you need to apply by filling out an application form. The community also has an Operator tool based on Airtable and Zapier for NoCode automation management, documentation, monitoring, and debugging. This is a great case study for a stable community development around an idea.

Makerpad Community is a community of software and automation developers, now based out of the Zapier facilities. The site features a board with various materials organized into groups: tutorials on using NoCode tools, website links, and job postings. The project used to have a forum, but that was shut down in 2022. Courses that used to be based on Makerpad have been moved to Zapier Learn.

Product NoCode Communities

Communities that focus on the use of specific products – primarily industry-specific solutions, such as a platform that integrates various NoCode and LowCode tools or a product that solves a large problem on its own – are essential.

Almost all popular products have their own communities with active forums: the NoCode integration platform Make, the tabular tool Airtable (59K members), the web builder Bubble, the mobile application builder Adalo, and others. And interaction with Notion's extensive community is, of course, structured through Notion itself.

A prime example of a strong product community is the Webflow Community. In particular, the product has a public backlog Webflow Wishlist – a platform where users can request and vote for new features. This allows the product team to better prioritize further development, and community members to contribute their ideas and evaluations.

In addition, some communities support using a specific group of tools. For example, NCF (No Code Founders) is excellent for those who actively use various NoCode tools and would like to save money on them. There is almost nothing of value in the basic free plan: you can create a profile and describe your startup. But in the Full Membership plan, for a membership fee of $99, you get not only access to the community's Slack channel but also $30k credits for the most popular NoCode tools: $500 Airtable credits, $500 Bubble credits for three months for new users, six months of free Plus plan from Notion, etc. There's also a $199 All Access option that gives you access to community courses.

Educational NoCode Communities

Almost all NoCode communities have educational features: tutorials, FAQs, and courses on various NoCode tools. However, some are specialized and built around an educational platform.

No Code MBA is a resource where you can learn the basics of a particular tool, such as Webflow, Figma, Bubble, Glide, Airtable, Zapier, or Google Sheets, in a few hours. In addition, the site contains interviews, podcasts, and tool picks, as well as complete guides to various no-code tools, making it a comprehensive resource for learning and development in the no-code field. NoCode MBA also provides hands-on project ideas and assignments, such as building chatbots, image generators, and content management systems.

100 Days Of No-Code is positioned as a campus for "creators, career switchers, and professionals" who are starting to use NoCode technology to launch products. It's easy for those who like to understand the issue by taking a course. There are both free, pre-prepared courses and paid courses with limited enrollment.

Nocode HQ offers step-by-step tutorials, NoCode templates, and expert sessions. A paid subscription gives access to all tutorials and four templates per month. Judging by the free version, the main focus is mobile apps and landings. The forum is closed.

WeAreNoCode is an online institute for product creators without a technical background, offering a virtual community and weekly group coaching sessions.

HR NoCode communities

Codemap is a meeting place for freelancers, agencies, and clients looking to hire NoCode experts. The site makes it easy to hire experts for software development, process automation, and consulting, which is ideal for projects that require rapid implementation and diversity in skills and technology.

The platform offers a large pool of experts skilled in various technologies. The search process starts with filling out a form, and then Codemap guarantees that it will find the right expert and that the project will begin within 48 hours.

WeLoveNoCode also provides services for recruiting and hiring proven NoCode experts for projects of varying complexity, from mobile app development to automation and MVP creation. WeLoveNoCode's specialty is using machine learning algorithms to select the right candidates.

Entrepreneurial communities with NoCode subsections

While specialized NoCode communities are a great source of knowledge, it's important to remember that such tools are only part of the bigger picture of developing and scaling startups. Looking at entrepreneurship, we will discover many valuable insights beyond the narrow NoCode niche.

For example, Indie Hackers is a community and platform for individual product developers, entrepreneurs, and startups to share ideas, strategies, and experiences in creating and developing online businesses. Approaches with and without no-code tools are discussed here. Participants share stories of successes and failures, discuss monetization, customer acquisition, and product optimization. It's a place to develop entrepreneurial savvy, find inspiration, practical advice, support the community, and learn about technology trends.

Product Hunt is a valuable resource for anyone interested in the latest trends in the world of technology and provides a platform to share ideas, get feedback, networking with professionals. Some users present new technology products, including apps, tools, and hardware, while others evaluate these solutions. The platform often hosts interviews with startup founders, discussions, and AMAs (Ask Me Anything sessions) with well-known entrepreneurs and technology experts.

If you use Twitter, subscribe to NoCode Influencers:

• @rrrhoover – Ryan Hoover, founder of Product Hunt
• @levelsio is an entrepreneur who has founded 5 successful NoCode startups (there is also a Telegram forum @levelsio)
• @sarahdoody – Sarah Doody is a renowned UX expert who has a lot of ideas for effective user experience for your product
• @bentossell – Ben Tossell, founder of Makerpad, who now focuses on AI tools
• @karenxcheng – Karen X. Cheng, known for applying NoCode and AI solutions to creative challenges
• @TaraReed_ – Tara Reed, a former Microsoft employee known for the Apps Without Code project, helps noobs build apps and launch businesses

An easy way to prioritize the right news stream is to explore Twitter listings options on NoCode.

Conclusion

The NoCode community is unusual in that technical issues are often resolved faster through collective intelligence and sharing of experience among developers than through documentation. Thanks to close ties, participants promptly share tooling tips and tricks. At the same time, the community's openness leads to the rapid spread of innovations within the industry – any developments are instantly available to all participants.

In addition, NoCode demonstrates an interesting "bottom-up" model of technology development – developers independently formulate requests for new solutions based on the needs of specific projects. This makes it possible to create tools that are really in demand.

Thus, integration with the NoCode community is not just a helpful bonus for a developer or entrepreneur but also a key to the success of a project based on NoCode or LowCode technologies. Unlike traditional development, where processes are closed (except for Open Source solutions), the NoCode community is generally more open and extroverted. This leads to a high feedback rate, product improvement, and, ultimately, faster time-to-market of innovative solutions. This synergistic effect of an open community is a unique competitive advantage of the NoCode approach.