Sam Altman, CEO of OpenAI, has raised an urgent concern about artificial intelligence that goes beyond job displacement or misinformation: cognitive atrophy, the gradual weakening of our thinking abilities when we outsource mental work to machines. The question he poses is deceptively simple yet profound: "How are we going to use these tools and make sure that we are stretching our brains more and more and continuing to understand the stuff that really matters?"

Sam Altman, CEO of OpenAI

This concern is not merely philosophical. A 2025 study conducted by researchers at MIT examined what happens in the human brain when people rely on AI to complete cognitive tasks. The study involved more than 50 participants who were asked to write essays using one of three approaches: ChatGPT, Google Search, or no digital tools at all. Researchers then measured brain activity, memory recall, essay quality, and how much ownership participants felt over their work.

The results were striking. People who wrote without any digital tools showed the strongest and most widespread connectivity across their brains. Those who used ChatGPT, by contrast, showed the weakest brain connectivity and had significantly more difficulty remembering what they had written. While the MIT study was not large enough to draw sweeping conclusions, the findings raise a critical question about what we lose when we stop practicing the kinds of thinking that AI can now do for us.

What Exactly Is Cognitive Atrophy and Why Should You Care?

Cognitive atrophy is the decline in mental capacity that occurs when we stop using certain thinking skills regularly. The brain, like a muscle, strengthens through use and weakens through disuse. Neuroscientists who study neuroplasticity have long understood this principle: the more we practice something repeatedly, the stronger and more efficient the neural circuits involved become. Conversely, the more we rest or stop using certain abilities, the more likely we are to lose them.

Consider a familiar example: GPS and online navigation tools. Many people have noticed that their ability to read maps and navigate by memory has weakened since they began relying on digital directions. The same principle applies to AI and thinking. If we use AI to write the first draft, summarize research, identify arguments, propose solutions, and anticipate objections, we need to ask what experiences we are no longer having and whether those experiences are precisely how good judgment develops in the first place.

This concern becomes especially acute when considering young people entering the workforce. A seasoned leader has accumulated thousands of experiences that inform their thinking: proposals they struggled to write, conflicts they navigated, arguments they made and later reconsidered, and real-life decisions that turned out well or failed entirely. These struggles create not just intelligence, but wisdom. Intelligence gives us the ability to learn, reason, and process information. Wisdom comes from having lived long enough and paid enough attention to know what to do with what we know.

How to Preserve Your Thinking Skills While Using AI Tools

The challenge is not whether to use AI, but how to use it in ways that preserve and strengthen our cognitive abilities rather than atrophy them. Experts suggest several approaches:

• Use AI to spark curiosity, not replace thinking: Rather than using AI to avoid thinking or to depend on machines for original conclusions, use it to find evidence for your own hypothesis or to explore questions you are genuinely curious about. This approach can actually grow your intelligence rather than diminish it.
• Preserve the struggle in important thinking tasks: Not all mental effort is created equal. There is value in wrestling with a problem, following your curiosity, making unexpected connections, developing a point of view, and learning to trust your own judgment. These are the struggles that build capacity for future challenges.
• Practice judgment and discernment deliberately: Over time, leaders develop many ways of knowing. They begin to sense when something is off, when an argument sounds convincing but does not quite hold together, or when a strategy looks good on paper but experience tells them there is another question worth asking. These capacities do not appear from nowhere; they are built through years of paying attention, making decisions, experiencing consequences, and learning from what happens next.

Daniel Goleman, author of the international bestseller "Emotional Intelligence," offers a useful analogy. Nobody goes to the gym because lifting something heavy is the most efficient way to move it from one place to another. We lift the weight because resistance is what develops strength. As AI becomes part of everyday work, leaders may need to become equally intentional about preserving the forms of cognitive resistance that develop human strength.

The Difference Between Efficiency and Development

In a culture that loves efficiency and optimization, the assumption is often that reducing mental effort is inherently good. After all, why spend precious hours doing work that a machine can accomplish in minutes, even if that work is good exercise for the brain? But this logic misses something important about how humans learn and grow.

Consider learning a language. Looking up a word is much easier than trying to remember it, but trying to remember is part of how we actually learn. The same principle applies to working through a complicated problem, struggling to verbalize an idea, contemplating a question when we do not immediately know the answer, or discovering halfway through an argument that what we thought we believed no longer quite holds up. The struggle itself is part of what builds our capacity to think.

This distinction matters for organizations and leaders. The question is not simply how or where AI might make people more productive. Leaders also need to consider what their people need to keep practicing in order to develop the capacities the future of their organizations will call for. There are many things we should be delighted to hand over to AI, particularly work that consumes time without developing much human capability. But there may be other forms of thinking we need to protect.

"How are we going to use these tools and make sure that we are stretching our brains more and more and continuing to understand the stuff that really matters," asked Altman.

Sam Altman, CEO of OpenAI

The goal, according to experts, is to discern which efforts are unnecessarily difficult and which are difficult because they are doing something important by building and strengthening capacities we will need now and later down the line. This is not because we want work to be harder, but because some struggle may be the only thing that prepares us for the work that lies ahead.