AI Is Rewiring How We Learn — and Why That Worries Neuroscientists

Generative AI is increasingly being marketed as a learning tool, but a new critique argues that the technology may be doing more than helping students save time: it could be weakening the very habits and institutions that build human thinking. The warning lands as schools, universities and major tech companies race to normalize chatbots in classrooms, even as evidence grows that heavy AI use can reduce effort, judgment and retention.

That is the central claim in a detailed essay published on October 5, 2026, which contends that human minds are not built like computers and should not be treated that way. The piece argues that AI is not simply another productivity tool, but a “cognitive shortcut” that risks turning education into a system of delegation rather than development.

At stake is more than a debate over homework. The broader question is whether schools, colleges and employers are optimizing for convenience at the expense of the long-term abilities people need to reason, remember, verify and decide for themselves.

Why the “brain as computer” idea is under fire

The article takes aim at a long-running metaphor in science and technology: the idea that the human brain works like a computer. That model, rooted in the work of Alan Turing and later amplified by computer scientists and AI leaders, imagines thought as a sequence of input, processing and output. It has shaped everything from neuroscience research to the design of machine learning systems.

But the essay argues that the model is incomplete. Human cognition, it says, is not just about internal calculation. It is deeply tied to bodies, movement, environment and social life. Rather than being passive processors of information, people constantly act on the world in ways that change what they perceive next.

That distinction matters because much of AI is built on a simplified version of cognition: feed in information, generate a response, produce an output. The critique says that view flatters machines and shrinks the complexity of human learning into something easier for software to imitate.

What is the alternative to the computational model?

The alternative, according to the essay, is to understand the brain as a feedback-control system. In this view, human action is not the final step after thinking; action is part of thinking itself. People move to influence the information they receive, and the nervous system evolves to help them do that efficiently.

That approach draws on the work of neuroscientist Paul Cisek, who argues that the brain developed through millions of years of interaction with changing environments. The point is not that the brain has nothing in common with computers. Rather, it is that a computational analogy misses the way perception, motion and decision-making are interwoven.

To make that concrete, the essay uses a classic baseball example. Catching a fly ball may look like a complex mental calculation, but in practice an outfielder relies on continuous adjustment: keep the ball in view, move to preserve that visual relationship, and let the body make constant corrections. The body is not solving a classroom problem; it is managing a living system in motion.

How does evolution change the way we think about the brain?

It changes the story by making human cognition look less like a static machine and more like an accumulated biological response to real-world challenges. The essay argues that nervous systems did not appear fully formed as abstract information processors. They developed over time as animals gained mobility, sensory range and new ways to control their surroundings.

As species moved from simple organisms to vertebrates, mammals and primates, new behaviors emerged alongside new neural capacities. One example discussed in the article is the hippocampus, which is associated with navigation and episodic memory. The point is not just that the brain stores data; it is that it helps organisms keep track of where they are, what happened before and what they might do next.

Seen this way, cognition is not merely internal computation. It is an extension of biological control into the world. Humans think in order to navigate, survive, coordinate and act with greater agency.

Why does that matter for AI?

Because AI systems are often sold as if they were the natural next step in the evolution of intelligence, when they may actually be something very different: tools optimized for substitution rather than development. The article argues that this substitution can feel efficient in the moment while eroding the deeper structures that make human learning durable.

That concern becomes sharper when AI is placed inside schools. If learning is understood as the formation of habits of mind over time, then a tool that repeatedly performs the thinking for students may interrupt the very process education is meant to support.

What makes AI dangerous to education?

The core worry is not that students will occasionally ask for help. It is that they will increasingly offload the hard work of thinking itself. The essay likens AI to junk food: appealing, easy and immediately satisfying, but harmful when it becomes a staple rather than an occasional indulgence.

In this analogy, the danger is cumulative. Just as a diet built around highly processed food can undermine physical health, a learning culture built around AI delegation can weaken intellectual stamina. Students may still finish assignments, but they may not build the knowledge structures that make later learning possible.

The article cites evidence that heavy chatbot use in schools can reduce engagement and shortcut the struggle that makes learning stick. It also points to studies suggesting that reliance on large language models may train users to trust answers too quickly and to apply less scrutiny elsewhere.

According to the article, AI should be understood less as a neutral tutor and more as a form of cognitive outsourcing that can gradually make independent reasoning less likely.

Which signs of harm are already visible?

The essay highlights several warning signs. Some students reportedly stop doing homework once AI becomes available. Others appear to do less original problem-solving and show weaker retention. Outside the classroom, concern is growing that people who lean heavily on chatbots may become more willing to accept output without verification.

There is also a social effect. If students see AI as the normal way to complete schoolwork, effortful thinking can start to look outdated rather than essential. That changes the culture of education, not just the mechanics of assignment completion.

• Less time spent wrestling with problems independently
• More reliance on machine-generated first drafts and answers
• Weaker habits of verification and critical judgment
• Greater temptation to use AI as the default solution

How are tech companies pushing AI into schools?

They are doing it by recasting chatbots as learning infrastructure. The article points to public remarks from OpenAI’s education leadership, which framed ChatGPT as a major learning platform and emphasized the scale of student use. Anthropic has also rolled out a teacher-focused initiative offering educators free access to its premium models.

These moves reflect a broader industry strategy: if AI can be normalized early in education, then future workers and consumers may become dependent on it by default. The article argues that the business logic is obvious even if the educational consequences are not.

In this telling, schools are becoming the next battleground for platform expansion. What is presented as support for teachers and students may also function as a long-term bid to make machine assistance inseparable from learning itself.

What is “cognitive delegation” and why does it matter?

Cognitive delegation is the practice of handing mental work to a machine when humans could do it themselves. The article cites researchers who warn that once people begin delegating reasoning tasks to AI, the surrounding environment can become less supportive of independent thought, which in turn makes future delegation even easier.

That creates a self-reinforcing loop. The more a person uses AI to think, the less practice they get thinking unaided. The less they practice, the more appealing AI becomes. The result is not total dependence overnight, but gradual erosion.

This is why the article argues for what it calls cognitive immunization: preserving habits and institutions that keep the mind active. Those include solo problem-solving, checking sources, arguing through disagreements and intentionally stepping away from AI when it is not necessary.

What role do schools play in resisting this trend?

Schools are central because they are one of the few institutions designed specifically to train independent cognition over many years. The essay says schools should be protected as spaces where students learn to struggle productively, not merely to arrive at correct answers quickly.

That is a direct challenge to any educational model that treats speed, convenience or personalization as the highest values. If learning is reduced to frictionless access, the article argues, then schools risk becoming delivery systems for machine assistance rather than places where judgment is built.

Why are some countries and districts pushing back?

Because the backlash is starting to look less ideological and more practical. The article notes that Norway has restricted AI use in schools for children under 13, while the American Federation of Teachers has urged limits on chatbots in elementary settings. Large districts including Los Angeles and New York City have also moved toward bans or strict guardrails.

These steps reflect growing concern that the youngest students may be especially vulnerable to dependency. At early ages, learning is not only about content; it is about forming the mental habits that underlie reading, writing, problem-solving and self-correction.

The article suggests that policy makers are beginning to recognize what product teams often do not: educational settings are not just markets for software, but ecosystems with long-term developmental consequences.

The essay argues that bans and restrictions are not anti-technology reflexes; they are protective measures meant to preserve the conditions under which human thinking develops.

How big is the resistance to AI in education?

It is growing, but unevenly. The article says public pushback is visible among students, graduates and some educators, including viral moments in which commencement speakers praising AI were booed by audiences. Some of that reaction likely reflects job anxiety. But the essay argues that it also reflects a deeper unease about losing one’s own intellectual development to machine shortcuts.

That sentiment is echoed by groups of students who frame their opposition to AI not as nostalgia, but as a defense of agency. The article highlights a student-led critique from Oberlin College’s Luddite Club, which described a commitment to self-actualization, imperfect effort and genuine inquiry rather than AI-assisted polish.

The resistance is important because it suggests the public debate is no longer only about efficiency. It is increasingly about identity, autonomy and what counts as meaningful learning.

What are students afraid of losing?

Students appear to fear more than bad grades or overreliance on tools. They fear losing the ability to know how they know what they know. If AI answers too much too quickly, the learner may miss the friction that produces understanding.

That fear is not limited to the classroom. In a world where writing, planning, coding and research can all be accelerated by AI, the question becomes whether people remain capable of sustained attention, original reasoning and intellectual resilience without a machine guiding every step.

The larger cultural argument: AI as a social technology

One of the article’s more important claims is that generative AI should be understood not as pure intelligence, but as a social and cultural technology. Rather than originating knowledge, it repackages patterns drawn from human work. In that sense, it depends on the accumulated insight of people who learned before it existed.

That framing shifts the debate. If AI is not a mind, but a system that distributes preexisting human knowledge, then the key question becomes what happens to the institutions that created that knowledge in the first place. Schools, libraries, peer review, apprenticeship, debate and careful writing all serve to develop and transmit understanding.

If those structures weaken, the article suggests, AI may appear to fill the gap while actually deepening the loss.

What does the article say we should do next?

It calls for restraint, especially in education. That does not mean banning every digital tool, but it does mean resisting the assumption that every friction point in learning should be removed. The article argues that the effort required to think is not a flaw in education; it is the mechanism by which education works.

From that perspective, AI should be limited to contexts where it supports human judgment rather than replaces it. The line is not always obvious, which is why the article urges stronger institutional safeguards and more deliberate public policy.

The practical implication is simple: if schools want students who can reason independently, they should not treat machine-generated answers as a substitute for the very work that develops reasoning.

Can AI ever help without causing harm?

Yes, but only if it remains subordinate to human learning goals. The article does not argue that every use of AI is damaging. Instead, it warns that routine reliance can crowd out the mental effort people need to build durable skill. Occasional assistance may be manageable; habitual dependence is the real risk.

That is the core of the “cognitive hot dog” metaphor. A small amount may not matter much, but a diet built around it changes the system. In education, the same logic applies to repeated AI use: the danger is not a single shortcut, but a culture of shortcuts.

Bottom line

The essay’s central warning is that AI adoption in education is outpacing our understanding of how human minds actually develop. If the brain is not a computer and learning is not mere information processing, then deploying chatbots as everyday academic companions may be a much bigger mistake than tech companies admit.

That makes the current policy debate unusually consequential. It is not just about whether students can use AI to write faster or teachers can use it to save time. It is about whether schools will remain places where people learn to think, or become places where thinking is increasingly outsourced.

For now, the pushback is real, the stakes are high, and the argument is only getting louder: if humans are more than machines, then education cannot be run as if they are not.