Do We Still Think? Or Just Ask AI?

My meeting just wrapped up, and soon after, I received the minutes. At first glance, they were impeccable. Well-structured, slightly lengthy but seemingly complete. But something felt... off.

As I read them more carefully, I realized they were less a summary of the meeting and more a faithful transcript. Every point was captured, yet the essence of the discussion was missing. The "why" behind the decisions had all disappeared. Everyone on the call seemed happy with the minutes. Ironically, when I later spoke to a few participants, none of them really remembered the discussions themselves.

Over the past few days, I've been finding myself increasingly uneasy with these perfectly polished outputs.

When I ask my team to prepare design documents, I often get architecture and framework-heavy designs that look perfect. They have all the world's best practices, reference all the right patterns, and are beautifully written. But what about their relevance? Very often, they have only a loose connection to the actual problem we're trying to solve. And when I ask, "Why did you recommend this approach?", many struggle to explain the reasoning behind what they've written.

That's what worries me.

AI has made it incredibly easy to produce great looking content. It's an amazing tool, and I wouldn't want to work without it. But somewhere along the way, are we beginning to mistake polished output for original thinking? Have we grown lazy and started delegating thinking?

Cognitive outsourcing

The success of AI assistants is no accident. They excel at precisely those tasks that humans find mentally demanding.

Throughout history, our greatest inventions have been attempts to overcome our limitations. When physical strength limited what we could build, we invented machines that could lift, transport and manufacture. When speed and accuracy of calculation became bottlenecks, we built computers that could process millions of operations in seconds. Every major technological leap has extended human capability.

Generative AI is the next step in that journey.

For the first time, we are building tools that can perform parts of what was once considered uniquely human cognitive work. They can read, summarise, write, translate, organise information, generate code, create presentations and even produce seemingly well-reasoned arguments. This is what I call cognitive outsourcing.

If AI starts doing all the thinking for us, what happens to our own?

Problem of Cognitive Outsourcing

The real risk of cognitive outsourcing is not that machines will think for us. It is that we may stop practising the kinds of thinking that make us uniquely human.

A new study from UC Berkeley analyzed more than 500,000 grades at a "large, selective public research university" in Texas. The pattern is hard to miss. In courses with a high share of writing and coding assignments, grades have climbed sharply since ChatGPT launched in November 2022. In a 2026 experimental study, researchers from Carnegie Mellon, MIT, Oxford and UCLA found that participants who used an AI assistant for as little as ten minutes were significantly more likely to give up or perform poorly once the AI was removed. The findings suggest that while AI can improve immediate performance, it may also reduce persistence and independent problem-solving when people become accustomed to relying on it.

The struggle that makes us uniquely human

Generating anything requires effort. Effort is of two types a) thinking and b) expressing the thought in the form of an artefact.  If we take a closer look at what insight really is and why it is so highly valued, we begin to appreciate the neuroscience that makes it possible. Just as exercising a muscle is important for its strengthening, the same is true in case of critical thinking. Thinking is like the brain gym. The skills most valuable to humans such as judgment, strategic thinking, pattern recognition, and decision-making under uncertainty are precisely the ones that develop through sustained cognitive effort. AI can accelerate the production of documents, but it cannot replace the mental exercise that builds these capabilities. Insight is not downloaded. It is built, one difficult thought at a time. If we consistently outsource that exercise, we may produce more work while becoming less capable thinkers.  We’re already seeing the results.

What can be done to protect cognitive fitness

Reward insights not outputs

Rewards reinforce behaviour. If instead of being blinded by polished looking sleek outputs we start asking for the why’s behind those outputs, teams will be forced to think. Leaders should ask What led you to this conclusion? What assumptions did you challenge? What alternatives did you consider? What changed your mind? These conversations reveal whether genuine insight exists or whether the work is simply a well-packaged synthesis generated by AI

Encourage verbal reasoning and writing by hand

Brainstorming ideas and encouraging verbal reasoning leads to original ideas and critical thinking. We encourage teams to explain their ideas without a presentation. This tells us whether they have just memorised the points or are they a result of genuine thought. There is growing evidence from cognitive science that writing by hand engages the brain differently from typing. It slows us down just enough to organize thoughts, make connections and remember information more deeply. Speaking and handwriting force us to construct thoughts rather than merely edit them. This is a capability that will be increasingly valued as AI tools proliferate.

Celebrate rough thinking and rough work

As a practice at Cere Labs, we handover a notebook to our team members right on the first day of their joining. The idea is that they take notes during their meetings and their first designs are always in a notebook. We’ve observed that the thinking doesn't always stop when the notebook closes. The problem continues to simmer in the background while they're walking, commuting or having lunch. Neuroscience refers to this as incubation, the brain's ability to continue processing information subconsciously after a period of focused effort. Many of our best ideas don't emerge while we're actively working; they appear later, when seemingly unrelated thoughts suddenly connect. A notebook doesn't create insight. But it creates the conditions in which insight is more likely to emerge.

Make meetings less about recording and more about thinking

AI meeting assistants have made it effortless to record, transcribe and summarize every discussion. While this is incredibly useful, it also changes how we participate. When we know that an AI is capturing everything, we subconsciously pay less attention, trusting that we can always "read the summary later." Recordings are invaluable for replaying a discussion. But deciding what truly mattered, what sparked a new idea, or what led to a decision, is still a fundamentally human act. AI should remember the conversation. Humans should remember its meaning.

Redefine AI’s role

As someone who leads an AI company, I will always welcome technologies that improve human lives. AI is one of the most transformative technologies of our time. The challenge isn't whether we should use it, but how we should define its role.

AI is an assistant, not a substitute for human cognition. It should take over repetitive, administrative and organizational work so that we can spend more time understanding problems, exercising judgment, questioning assumptions and generating original ideas.

When machines took over physical labour, we became physically less active. The result? We invented gyms to deliberately exercise muscles that our daily lives no longer demanded.

As AI takes over more cognitive work, we may soon need the equivalent of brain gyms, deliberate practices that keep our minds engaged in reading deeply, reasoning independently, debating ideas and creating insight.

The goal is not to compete with AI. It is to ensure that, while AI becomes smarter, humans continue to become wiser.

Concluding thoughts

The future will belong to people with insights.

There is an uncomfortable irony here. We know that deep thinking produces better decisions, yet the modern workplace rewards speed above all else. Quarterly targets, overflowing inboxes, back-to-back meetings and constant notifications leave very little room for reflection. AI didn't create this problem. It simply made it easier to prioritize speed over thought.

Perhaps we are asking the wrong question. Instead of asking, "How can AI save us time?", we should be asking, "What should we do with the time AI saves?"

AI is giving us back time. That is one of its greatest promises. But time, by itself, creates no value. What matters is how we use it.

If every minute AI saves is immediately consumed by more meetings, more emails and more output, we may become more productive without becoming wiser. The real opportunity is not just to work faster, but to think better.

Perhaps the true measure of AI's success will not be how much work it removes from our day, but whether it gives us back the time to do the kind of thinking that only humans can do. Our capacity to think deeply has been humanity's greatest evolutionary advantage. AI should strengthen that advantage, not slowly erode it.