In the shadow of the intelligent machine: the construction of the obedient mind

When discussing a technology, we often ask what it is; yet the more pressing question is for whom, by whom, and to reproduce which social order that technology was designed. The debate over artificial intelligence begins precisely here. This new tool, which permeates every sphere from artworks to Hollywood studios, from university lecture halls to office buildings, is presented as if it were merely a tool for efficiency.

It saves time, reduces costs, and makes life easier. All of this is true. But it’s incomplete. Because, unlike the tools that came before it, artificial intelligence is a mechanism that can learn, renew itself, and influence decision-making processes. For the first time in history, a tool does not merely transform its user; it seeks to think and make choices on their behalf.

Therefore, discussing the issue solely within the “beneficial or harmful” dichotomy is insufficient. The true nature of artificial intelligence cannot be separated from the production relations into which it is embedded. The thesis defended in this article is simple: What we call artificial intelligence today cannot be evaluated as merely a technological breakthrough. It is a new regime of capital accumulation, and this regime carries the risk of producing a compliant type of subject while redistributing knowledge, labor, and even thought.

The question of the owner, not the tool

The massive data centers training AI models, the multi-billion-dollar chip infrastructure, and the energy grids are not randomly scattered across the globe. The companies that build, train, and lease these models can be counted on one hand. This concentration is not a technical choice but a structural reality. Because data—the raw material of artificial intelligence—only generates value as scale increases; and as scale increases, only very large capital groups can remain in the game. Although we see small investments here and there, these are merely those adapting the infrastructure provided by the big tech giants to their own needs.

At this point, we need to flip the concept on its head: Artificial intelligence is not a tool, but a form of property. Everyone who uses it is, in fact, connected to the infrastructure of a specific property owner. Every query that appears to be free provides data in return. Every piece of data makes the model more valuable; every increase in value flows to where ownership is concentrated. In classical capitalism, the worker sold their labor. Today, the user—often without realizing it—donates their cognitive labor and attention.

As Yanis Varoufakis puts it, every time we go online to use algorithmic services, we have no choice but to make a Faustian bargain with their owners. We submit to a business model based on the collection of our data, the tracking of our activities, and the invisible curation of our content—all to use the personalized services provided by algorithms… We are becoming free servants who provide behavioral patterns that predict our actions, guide our preferences, influence our decisions, change our minds, and train our attention[1].

The price of convenience: cognitive debt

Capital’s historical success stems not only from its control over the means of production but also from its ability to dominate the habits of the producers themselves. If you start referring to a need by the brand’s name, that is the triumph of the investment[2]. Artificial intelligence is opening up an entirely new front right here, infiltrating not only everyday language but also the practices of daily life. As it draws from more sources, we need to think more deeply about the impact it creates on its users. It has been nearly three years since Silicon Valley began aggressively marketing large language model-based chatbots like ChatGPT as the inevitable future of everything, and the group feeling this pressure the most has been Generation Z. We are beginning to see the results of research conducted on this topic.

A recent study at the MIT Media Lab showed that students who relied on AI during their research exhibited a significant drop in brain activity measured by EEG[3]. Fifty-four undergraduate students were randomly assigned to write an essay using AI, a search engine, or on their own, and EEG scans were conducted during this time to measure the electrical activity in their brains. A decrease in brain activity was observed among those who wrote with AI, and most students who used AI could not even quote a single sentence verbatim from what they had written. In fact, for four months, AI users consistently demonstrated lower performance at neural, linguistic, and behavioral levels.

Researchers gave this phenomenon a striking name: cognitive debt[4]. The convenience gained today is accumulating as a debt that will be paid back with interest tomorrow. AI boosts performance in the short term, but in the long term, it erodes determination, perseverance, and the ability to solve problems independently. When people switch to working without assistance, they struggle. This is because the brain is deprived of the experience of “overcoming a challenge on its own.”

Two more concepts have been added to this in the literature: cognitive atrophy and cognitive surrender. Approximately two-thirds of young adults now access information not through search engines, but directly through an AI assistant. According to a recent Harvard-Gallup study[5], 74% of the young people surveyed (those born between 1997 and 2012) say they use a chatbot at least once a month. (Another study found that more than half of college students in the U.S. use these tools weekly for their studies.)

Generation Z is concerned that the use of AI will eliminate three key applications: learning by doing, critical thinking, and learning from others. AI is taking over the skill of learning by doing or replacing the process of learning from people, including peers and mentors. However, most importantly, it prevents deep or critical engagement with ideas and information.

Yet a significant portion of the summaries produced by AI assistants—between 10% and 28% according to some estimates—are flawed or biased. When you consider the trillions of queries processed annually, the scale of misinformation circulating unchecked becomes clear. More importantly, we are outsourcing not just information, but the very act of thinking. We accept AI results “as is” without criticism.

This phenomenon is not innocent from a political economy perspective. Because a subject who does not criticize, does not question, and does not filter what they read through their own lens is the most compliant audience for any form of power. Critical reason, a fundamental achievement of Enlightenment thought, has never existed spontaneously in history. It was the product of a long practice of reading, writing, debating, and making mistakes. Now, the very foundation of this practice—pausing, going back, searching for the right word, making mistakes and correcting them—is being quietly sidelined. Even research in the previous technological generation, based on the repetition of another text via cut-and-paste, served learning indirectly. Because to choose what to quote, you had to at least read and sift through what you had read. Today, even that is no longer necessary.

Instead of a mind capable of critical and philosophical thought, a narrow “instrumental mind” emerges—one that accepts the speed and functionality offered by technological tools and focuses on choosing the most practical path to achieve a goal. This enables algorithms and data analytics to steer human behavior, beliefs, and decisions. Today, this process manages individuals to remain within the boundaries set by the system, even as they believe they are acting of their own free will.

A Critique of the relationship, not the tool

“The less you understand something, the more respect you show it, and the more you bow down before it.” [6]

My intention in saying all this is not to disparage artificial intelligence. The issue is not the tool itself, but the relationship we establish with it and the power dynamics within which this relationship is embedded. What makes a technology political is not the silicon it contains, but the regimes of ownership, labor, and knowledge through which it operates.

That is why we must think on two levels at once. At the individual level, we must maintain the distinction between using artificial intelligence as an assistant and gradually delegating the act of thinking to it. The brain is also a muscle. It atrophies where it bears no weight. At the societal level, the issue is far broader. Who is training these models, with what data, under whose oversight, and in the interest of what public good? Who does it see? Who does it exclude? Leaving the future of artificial intelligence to the boardrooms of a few companies could be democracy’s quietest loss over the next decade.

In the shadow of the intelligent machine, the core question remains the same: What are we feeding ourselves? Are we raising a generation that is adept at following instructions but incapable of independent thought? Drawing on Foucault, perhaps the illusion that we are free in our choices is a revelation of just how controllable we are. Let us use artificial intelligence, but let us not cede to it the right to choose how we think—and thereby how we are governed.

The counter-movement

Today, a large portion of young people and students—who are most directly affected by artificial intelligence—feel deep anger and even resentment toward an AI-driven future they perceive as being forced upon them.

While young people’s interest in AI persists, skepticism is growing. According to the findings of a Gallup survey conducted with a sample of 1,572 people[7], even the most active users of AI are less positive about it compared to a year ago (2025). Generation Z is less convinced than in 2025 that AI enhances learning and task completion efficiency. The percentage of Gen Z respondents who agree or strongly agree that AI tools can help speed up work has dropped by 10 points compared to 2025, while consensus on AI’s ability to accelerate learning has decreased by seven points, falling to 46%.

The number of Gen Z employees who believe the risks of AI outweigh its benefits has also increased by 11 percentage points compared to last year, reaching nearly 50%. And while 56% say these tools help them finish their work faster, 8 out of 10 acknowledge that using AI in this way will make genuine learning even more difficult in the future. Gen Z workers place more trust in work completed without AI (69%) than in AI-assisted work (28%).

The younger generation in the workforce is far more aware of the losses. The sentiment that we will develop our own skills and productive labor rather than strengthen Silicon Valley is spreading. This movement is also growing among science workers and academics. For example, a group led by cognitive scientists and AI researchers from universities in the Netherlands, Denmark, Germany, and the U.S. published a strongly worded statement calling on educators and administrators to reject corporate AI products. In the statement titled “Against the Uncritical Adoption of 'AI' Technologies in Academia,” the authors write, “When it comes to the AI technology industry, we reject their frameworks, reject their addictive and fragile technologies, and demand the restoration of the sanctity of the university both as an institution and as a set of values,” they write[8]. Another example is France. One of France’s leading universities, Sciences Po, has banned students from using ChatGPT for any assignments or presentations. For my part, I have adopted a similar approach for my own course by establishing a workflow that evaluates PDF-based articles.

There are people who worry that artificial intelligence is a constantly evolving element that replaces human labor, human thought, and human-to-human interaction. Research shows us that these concerns are certainly valid. The difference between using artificial intelligence as a tool and quietly handing over the task of thinking to it is as much a matter of individual attention as it is a matter of political stance. Who is training these models? With what data? Under whose oversight? As long as these questions remain unasked, technological progress will amount to nothing more than a redistribution of power. Critical reason has never existed spontaneously in history; it was the product of a long practice of reading, writing, making mistakes, and correcting them. Maintaining that practice is the freest action still possible in the shadow of the machine. (ÖB/EMK/VK)

References

[1] Yanis Varoufakis, Technofeudalism: What Killed Capitalism, Trans. Mustafa Güdük, Diplomat Yayınları, (2026)

[2] We can say this about the influence of brands like Coca-Cola and Selpak on language.

[3] https://arxiv.org/abs/2506.08872

[4] https://www.media.mit.edu/publications/your-brain-on-chatgpt/

[5] https://hbsp.harvard.edu/inspiring-minds/how-gen-z-is-using-ai

[6] Wilhelm Reich, Listen, Little Man!, Cem Yayınevi (2022)

[7] https://news.gallup.com/poll/708224/gen-adoption-steady-skepticism-climbs.aspx

[8] Against the Uncritical Adoption of ‘AI’ Technologies in Academia

 https://www.bloodinthemachine.com/p/cognitive-scientists-and-ai-researchers, https://zenodo.org/records/17065099

https://www.reuters.com/technology/top-french-university-bans-use-chatgpt-prevent-plagiarism-2023-01-27/