Is AI Reducing Our Ability to Think?

Is AI reducing our ability to think?

How Artificial Intelligence Is Changing the Way We Learn

"Education is not the filling of a pail, but the lighting of a fire." Often attributed to William Butler Yeats, though the line is widely believed to be misattributed, its likely origin traces back to Plutarch. It seems a fitting note to strike at the outset of a piece about the value of verifying what we accept as true.

Every time we genuinely struggle to understand a new concept, solve a difficult problem, or explain an idea in our own words, something remarkable happens inside our brain.

The human brain contains approximately 86 billion neurons, connected through an extraordinarily complex network of synapses. Every meaningful learning experience slightly reorganises this network, strengthening some connections while weakening others, a remarkable property known as neuroplasticity.

Modern AI systems are also built from networks of interconnected computational units inspired by biological neurons. But there is one fundamental difference. Human beings learn largely as individuals, sharing knowledge slowly through teaching, writing, and conversation. AI systems, by contrast, can transfer what one model has learned almost instantly to millions of identical copies. As AI pioneer Geoffrey Hinton has often observed, this ability to share learned representations gives digital intelligence an unprecedented speed of collective learning.

Yet speed is not the same as understanding. Human learning is shaped by experience, emotion, curiosity, judgement, and the effort of making sense of the world. The question is not whether AI can learn faster than we do in certain domains, it clearly can. The more important question is whether our increasing reliance on AI changes how we learn.

This remarkable ability, known as neuroplasticity, is one of the defining characteristics of human intelligence. It explains why children learn languages effortlessly, why adults can master new skills, and why repeated practice gradually transforms confusion into expertise.

Yet this understanding raises an important question.

If Artificial Intelligence increasingly removes the struggle from learning, what exactly happens to this process?

This is no longer a philosophical question. It is becoming the subject of serious investigation by neuroscientists, cognitive psychologists, education researchers, and AI scientists. While the research is still in its early stages, it points toward a concern that deserves the attention of every student, parent, teacher, and professional.

The issue is not whether AI makes us less intelligent.

The issue is whether it quietly encourages us to skip the very mental effort that makes learning possible.

Why Struggle Is Essential for Learning

For decades, cognitive psychologists have studied an idea that initially sounds counterintuitive.

Learning that feels more difficult often produces better long-term results.

Researchers Robert Bjork and Elizabeth Bjork describe this phenomenon as desirable difficulties. When learners retrieve information from memory, solve challenging problems, or generate their own explanations instead of simply reading them, they retain knowledge longer and apply it more effectively in unfamiliar situations.

In other words, the struggle is not an unfortunate side effect of learning. The struggle is the learning.

Anyone who has learned to ride a bicycle, solve mathematical equations, play a musical instrument, or speak a new language has experienced this process.

The first attempts are frustrating.

Mistakes are frequent.

Progress feels slow.

Yet each attempt strengthens understanding in ways that passive observation never can.

Educational researchers have repeatedly shown that techniques such as retrieval practice, spaced repetition, self-explanation, and problem solving produce stronger and more durable learning than simply reviewing correct answers. The brain remembers what it has worked to understand.

AI Changes the Learning Equation

For centuries, education struggled with one fundamental limitation: access to information.

Books were scarce.

Teachers were limited.

Libraries were distant.

Today, the opposite is true. Information has become almost unlimited. With tools such as ChatGPT, Claude, Gemini, Microsoft Copilot, and Perplexity, answers to almost any question can be generated within seconds.

This is an extraordinary achievement.

Students can receive personalised explanations.

Professionals can accelerate research.

Teachers can prepare better lesson plans.

Researchers can explore ideas more rapidly than ever before.

The productivity benefits are undeniable. But every technological revolution changes not only what we can do, but also what we choose not to do ourselves.

The calculator reduced the need for routine arithmetic.

GPS reduced our reliance on navigation skills.

Search engines reduced the need to memorise facts.

Generative AI may reduce something even more fundamental.

It reduces the need to wrestle with ideas before arriving at an answer.

What the Research Is Beginning to Show

In 2025, researchers at the MIT Media Lab published a preliminary study examining how people wrote essays using generative AI.

Participants who relied heavily on AI showed lower levels of neural engagement during the writing task and demonstrated weaker recall of what they had written compared with participants who completed the work independently.

The study involved a relatively small sample and should not be interpreted as definitive proof of long-term effects. Like all early research, it requires replication across larger and more diverse populations.

Nevertheless, it raises an important possibility.

When AI performs much of the cognitive work, our own brains may simply engage less deeply with the material.

This observation aligns with decades of established research in cognitive psychology.

Learning becomes durable not because we have seen the correct answer.

It becomes durable because we have actively constructed understanding ourselves.

The MIT findings therefore do not stand alone.

They fit within a much broader body of evidence about how human learning has always worked.

Three Risks We Should Be Paying Attention To

1. Reduced Cognitive Engagement

Learning requires attention.

It requires making connections, testing assumptions, recognising errors, and reorganising existing knowledge.

When AI performs these activities on our behalf, we may still obtain accurate answers, but we engage less deeply in the thinking that produced them.

Efficiency increases.

Understanding may not.

2. Critical Thinking Atrophy

   
   

Perhaps the greater danger is not incorrect answers.

It is unquestioned answers.

Generative AI is remarkably good most of the time.

Because it is usually helpful, users gradually develop the habit of accepting its responses with minimal scrutiny.

Yet in professional life, unquestioned assumptions often create the biggest mistakes.

An engineer cannot afford to overlook an incorrect calculation.

A doctor cannot accept unsupported medical advice.

A lawyer cannot rely on fabricated case citations.

A business analyst cannot ignore inconsistent data.

The skill most at risk is therefore not memory.

It is verification.

In the AI era, the ability to question, verify, and evaluate evidence may become one of the most valuable professional competencies.

3. Compounding Skill Debt

Students face an additional challenge that professionals often do not.

Education is cumulative. Every concept becomes the foundation for another.

A student who consistently asks AI to solve algebra problems may successfully complete today's homework.

But algebra is the language of calculus.

Calculus supports physics, engineering, economics, statistics, data science, and machine learning.

Missing one layer weakens every layer that follows. I think of this as Compounding Skill Debt. Just as financial debt accumulates interest over time, learning debt accumulates intellectual cost. Each concept that is bypassed today increases the difficulty of understanding tomorrow's concepts.

Eventually, students are no longer struggling because the subject is inherently difficult.

They are struggling because the foundations were never fully built.

AI Is Not the Enemy

None of this means students should avoid AI. That would be both unrealistic and undesirable. Artificial Intelligence is one of the most transformative technologies ever developed. Students who refuse to learn it will be at a disadvantage.

The goal is therefore not to reject AI. It is to use AI in ways that complement how the human brain learns rather than replace it. The distinction matters enormously.

Seven Principles for Learning with AI

Students, educators, and professionals can benefit enormously from AI by adopting a few simple habits.

Think before prompting. Spend a few minutes trying to solve the problem yourself before asking AI.

Use AI as a tutor, not a substitute. Ask for explanations instead of ready-made answers whenever possible.

Challenge the response. Request alternative viewpoints, assumptions, and counterarguments.

Verify important information. Never assume AI is correct simply because it sounds convincing.

Explain concepts in your own words. If you cannot explain an idea without AI, you probably do not understand it yet.

Iterate instead of copying. Treat AI's first response as the beginning of a conversation rather than the finished product.

Protect productive struggle. Do not allow AI to remove every difficult step. Some struggles are precisely what make learning durable.

The Real Question

We often ask whether Artificial Intelligence is becoming smarter.

Perhaps the more important question is whether we are continuing to exercise the habits that make human intelligence possible.

Writing did not make humanity forget how to think.

Calculators did not destroy mathematics.

Search engines did not eliminate knowledge.

Likewise, AI will not make us intellectually weaker simply because it exists.

But if we allow it to replace curiosity with convenience, verification with acceptance, and learning with copying, we risk weakening capabilities that no technology can fully replace. The future will not belong to those who avoid AI. Nor will it belong to those who delegate all thinking to AI.

It will belong to those who understand when to leverage artificial intelligence and when to engage their own intelligence first.

In an age where information has become abundant, the ability to think independently may become humanity's greatest competitive advantage.

Note on the Evidence

The MIT Media Lab study cited above is an important early contribution to understanding how generative AI may influence learning and cognitive engagement. However, it involved a relatively small sample and should be interpreted as preliminary evidence rather than definitive proof. Its findings nevertheless raise important questions that deserve further investigation alongside decades of established research in neuroscience and cognitive psychology.

About India Career Centre

At India Career Centre, we believe the purpose of education is not merely to produce answers, but to develop capable thinkers who can ask better questions, exercise sound judgment, and continue learning throughout their lives. AI can accelerate that journey—but it should never replace it.

"I have, in the past few years, used LLMs to learn, research, and write about various subjects. I have personally found the tool very useful. It is just that, along with that, one should not lose the ability to think originally about any subject. One should also not lose the ability to be always curious about everything they observe around them, and to question the status quo. Only then can something new be developed."

Dr. Sp Mishra, Founder, India Career Centre

References & Further Reading

• Kosmyna, N., et al. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. MIT Media Lab.

• Kosmyna, N., et al. (2025). Your Brain on ChatGPT. arXiv Preprint.

• Bjork, R. A., & Bjork, E. L. Desirable Difficulties in Learning.

• Brown, P. C., Roediger III, H. L., & McDaniel, M. A. Make It Stick: The Science of Successful Learning.

• Kahneman, D. Thinking, Fast and Slow.

• Dunlosky, J., et al. Improving Students' Learning With Effective Learning Techniques.

• National Academies of Sciences, Engineering, and Medicine. How People Learn II.