Thinking in the Age of AI: Replacement, Augmentation, or Loss? — Kafedra Analytics

Thinking in the Age of AI: Replacement, Augmentation, or Loss?

The question of whether artificial intelligence will replace humans has become a commonplace. It is discussed at all levels — from kitchen conversations to parliamentary hearings. However, over the years of these discussions, a stable optics has formed that may turn out to be not just incomplete, but fundamentally erroneous.

We propose to look at the problem differently. Not from the position of “will it replace — will it not replace,” but from an architectural position: what happens to thinking itself when it is embedded in an environment where AI has become not just a tool, but part of the cognitive circuit?

Part 1. Accumulation of Cognitive Debt: The Price of Delegation

In October 2025, researchers at the Massachusetts Institute of Technology published the results of an experiment that was meant to serve as a warning. It involved 54 people divided into three groups. One wrote essays using Google, the second used ChatGPT, and the third relied solely on their own brain. Brain activity was recorded using an EEG device [1].

The result was telling. Those who used ChatGPT demonstrated the least engagement in the work at the neural, linguistic, and behavioral levels. They didn’t just “think less” — they essentially did not identify the finished text with themselves and could not subsequently recall its content [1].

The authors of the study introduced the term “cognitive debt”: a situation where saving mental effort in the short term leads to their deficit in the long term, manifesting in a decline in critical thinking, vulnerability to manipulation, and a drop in creativity [1].

This phenomenon is reproduced in different contexts and by different research groups. According to a study by Gerlich (2025), frequent use of AI correlates with lower critical thinking scores [2]. The younger age group — from 17 to 25 years old — is particularly vulnerable to cognitive debt [2].

Political scientist Georgy Bovt formulates the problem extremely harshly: “Excessive trust in AI can lead to the loss of skills in independent information analysis. And then people will become ‘brainless’ slaves of AI” [3]. He draws a parallel with studies showing that pilots who rely on autopilot lose manual control skills; drivers who use navigation navigate worse without it [3]. Neural connections, like muscles, atrophy without practice.

Part 2. Replacement of Thinking or Its New Form?

However, a fundamental question arises here. Is AI simply a “replacement” for thinking, or is it restructuring its very architecture? The answer is not as simple as it seems.

In a study published in the journal AI and Ethics (2026), the authors arrive at a paradoxical conclusion: instead of creating machines that resemble humans, we are redefining humans in mechanistic terms [4]. The field of AI, in their view, does not so much create human intelligence in a machine as it redefines what it means to be human, doing so in terms of computability [4].

The authors call this process the “machinization of man.” They identify three dimensions of it: the replacement of human thinking with computation, the subordination of art and culture to computation, and the replacement of active decision-making with passive acceptance of statistical results [4]. Traditionally, religion helped cope with existential uncertainty by offering fixed ideas about human nature. Today, according to the researchers, AI takes on a similar role — freeing us from the burden of self-determination and not knowing what it means to be human [4].

The authors ask a different question: why do people even want to use technological tools to become less human? Their answer: because AI promises to free us from the “burdens of existence” — from the anxiety of not knowing, from the need to make decisions, from responsibility for self-determination [4].

Part 3. The KsA Model: Returning the Ability to Discern

The response to the challenges of AI may lie not in rejecting the technology, but in developing fundamentally different ways of interacting with it.

A paper published on Zenodo (2026) proposes the “KsA” model (Discernment — Fixation — Adaptation) [5]. Its authors proceed from the premise that the problem is not that AI “takes away” thinking, but that it completes it, leaving no phase for discernment — a moment of pause in which a thought could be born [5].

The authors of the model refer to cognitive science and neurophysiology. They argue that at the cognitive level, the use of AI is accompanied by a weakening of metacognitive control (Poldrack, 2023) [6], and at the neurophysiological level — by a decrease in the activity of the Default Mode Network, responsible for reflection and inner speech [7].

The KsA model tracks not what a person thinks, but how. Three axes are recorded: Discernment (does a pause, clarification, doubt arise), Fixation (does repetition, automatism intensify), and Adaptation (does a shift, reformulation occur). High R + active λ + falling I = thinking returns. Low R + high I + blocked λ = automatism [5].

Part 4. Intelligence as an Interface: A Philosophical Parallel

In the material “Intelligence as an Interface” (April 2026) [8], mentor Silicari Ajahary offers an alternative concept that deeply resonates with our analysis. He argues that thought is not born in the head but comes from outside, and the brain acts not as a generator, but as a receiver and decoder [8].

This model changes the entire optics of perceiving intelligence. If thought is a signal from the information field, then intelligence is the ability to connect to the signal, recognize its structure, assess applicability, and implement it in practice [8]. The author identifies four functional types of intelligence: receptive (captures new ideas), coordinating (finds those capable of implementing the idea), structuring (breaks down the task into stages, identifies dependencies) and executive (translates the plan into action) [8].

This model directly relates to the challenge that AI poses to us. If AI is a new “module” with colossal computing power but no built-in ethical compass, then the task of humans is not to compete with it in speed, but to learn how to formulate the right queries and preserve the right of final choice [8].

Part 5. Historical Parallel: The Philosophy of Krishnamurti

In 1980, when the 85-year-old philosopher and mystic Jiddu Krishnamurti became acquainted with the idea of artificial intelligence, he immediately recognized in it not a technological threat, but a philosophical and psychological challenge [9]. He was concerned not that machines would become like people, but that people already have machine-like minds [9].

Krishnamurti warned: an insufficiently cultivated mind, used exclusively for material and mechanical purposes, will be ideally imitated and, consequently, replaced by computers and machines [9]. His question was precise: “If a machine can do everything that a human can do, and do it better than us, then what is a human being?” [9].

This question is more relevant today than ever. And the answer to it, according to Krishnamurti, lies not in technology, but in the ability of a person to develop those qualities of mind that by definition cannot be mechanical: the ability to discern, to pause, to reflect.

Part 6. Architectural Conclusion: From Automatism to Discernment

AI is particularly strong in continuation: text, reasoning, strategy, chains of solutions. As mentor Silicari Ajahary notes, “we still do not fully understand its nature, but it is already obvious: its potential is super-scalable, and its speed of evolution is astonishing” [8]. AI has absorbed a colossal amount of data — everything that humanity has been posting online for decades: articles, dialogues, theories, mistakes, revelations, and everyday thoughts. “In essence, it has become a digital reflection of the collective consciousness” [8].

This is precisely why in the new architecture of thinking, the key ability of a person becomes not the ability to generate continuation, but the ability to interrupt automatism — to stop the predictable course of reasoning and switch to a mode of discernment: to see that there is always otherwise.

As precisely formulated in the KsA model, discernment is not the depth of thought and not originality. It is the ability not to confirm immediately [5]. It is the ability to ask oneself questions that interrupt inertia:

These questions do not lead to an answer — they return us to a state of choice. Unlike AI, which always moves forward along the most probable trajectory, a person can stop, turn around, and see that the solution may not necessarily have to be found within the framework of the proposed formulation.

Part 7. Facts, Not Forecasts

Study	Subject	Key Conclusion
MIT (2025) [1]	Cognitive activity when writing essays with AI	Lowest engagement when using ChatGPT, “cognitive debt” accumulation
Gerlich (2025) [2]	Relationship between AI use and critical thinking	Correlation between frequent AI use and lower critical thinking
“AI and Ethics” (2026) [4]	Machinization of man	AI does not create human intelligence but redefines the human in computable terms
KsA Model (2026) [5]	Phase diagnostics of thinking	Loss of the discernment phase between question and answer
“Intelligence as an Interface” (2026) [8]	Nature of intelligence and the challenge of AI	AI is a new module without a built-in ethical compass; the task of a person is to preserve the right of choice

Conclusion: Thinking as Discernment

The materials we have gathered and the studies we refer to allow us to draw the following conclusions:

The future does not lie in rejecting AI, but in a new quality of thinking. As Krishnamurti wrote, the question is not how to protect ourselves from machines, but how to develop within ourselves what cannot be reproduced by a machine [9]. And the answer to this question lies in the human ability to maintain the phase of discernment — the ability to stop, doubt, and see otherwise.

Sources

[1] MIT researchers found that heavy reliance on ChatGPT leads to lower cognitive engagement and “cognitive debt” when writing essays (October 2025).

[2] Gerlich, M. “AI Use and Critical Thinking Decline: Correlational Study” (2025).

[3] Bovt, G. “Excessive trust in AI can lead to the loss of skills in independent information analysis” // public speeches and comments (2026).

[4] “AI and Ethics” Journal (2026) — a study on the “machinization of man” and the redefinition of human nature in terms of computability.

[5] KsA Model (Discernment — Fixation — Adaptation) // Zenodo (2026). — proposes phase diagnostics of thinking in the age of AI.

[6] Poldrack, R. “Metacognitive Control and AI: Neurocognitive Perspectives” (2023).

[7] Default Mode Network activity and AI-mediated reflection — neurophysiological studies (2024–2026).

[8] Silicari Ajahary. “Intelligence as an Interface” // Gcalf.com (April 26, 2026) — the concept of intelligence as a signal receiver and the challenges of AI.

[9] Jiddu Krishnamurti. “The Future of Mind and Machine” // Discourses and Conversations (1980).

The material was prepared by the editorial board of the journals “Kafedra” and SforNews based on open data, research from MIT, Springer, IEEE, the KsA model, as well as the concept of “Intelligence as an Interface.” When citing, a reference to the original source is mandatory.