AI Is Changing How We Think. Here Is How to Stay in Charge
Most of us use AI the wrong way, warns David Rock. Here is what the research says, and what to do instead.
A familiar evening
It is 10pm and your report is due in the morning. The spreadsheet still makes no sense, so you paste it into a chatbot and ask for key insights. Ten minutes later, neat bullet points are sitting in your slides.
The next day, your manager asks a simple follow-up question about one of those insights. You freeze. You delivered the answer, but you never did the thinking.
If that scene feels familiar, you are in large company. A Google and Ipsos survey found 88 per cent of Nigerian adults have used an AI chatbot, against a 62 per cent global average. Some 93 per cent said they use AI to learn or understand complex topics.
The expert's warning
David Rock is co-founder and chief executive of the NeuroLeadership Institute. He wrote the book Your Brain at Work and is a keynote speaker at UN Skills Fest 2026. In a UN News feature published on 29 October, he argued that many of us are handing AI our thinking, not just our tasks.
Rock describes a shift from "AI FOMO", the rush to adopt generative AI, to an "AI hangover". Organisations raced to roll out the tools, he says, often before understanding the long-term effects. Now the consequences are becoming clearer.
His most striking estimate is that perhaps only 5 to 10 per cent of workers use AI as intended. That minority hands routine work to AI to free up mental capacity. Most people do the opposite, outsourcing analysis, evaluation and even decisions, which he warns erodes judgement.
He also sees a crisis of attention. Messages, notifications and AI-generated summaries pile up faster than our brains can process them. He likens it to the arrival of email, only on a far larger scale.
Then there is connection. Rock cites research suggesting that heavy users of AI companions may feel lonelier, not less lonely.
A four-week MIT Media Lab and OpenAI experiment with 981 participants points the same way. Heavier daily ChatGPT use went with more loneliness and less socialising, though the link does not prove cause.
His remedy is metacognition, which simply means thinking about your own thinking. Use AI to challenge and widen your view, he says, while people stay in charge of judgement. As he puts it: "When your brain's not in the game, you very quickly lose those skills."
What the research shows
Rock's warning matches a growing body of research, although the evidence is still young. Four studies stand out.
| Study | Year | Who took part | Key finding |
|---|---|---|---|
| MIT Media Lab, "Your Brain on ChatGPT" (preprint) | 2025 | 54 participants writing essays | ChatGPT users showed the weakest brain connectivity and struggled to quote their own essays |
| Microsoft Research and Carnegie Mellon University (CHI 2025) | 2025 | 319 knowledge workers, 936 real work examples | More confidence in AI went with less critical thinking; effort shifted to checking AI output |
| Bastani and colleagues, PNAS | 2025 | Nearly 1,000 high school maths students in Turkey | Open chatbot access lifted practice scores, but exam scores fell 17% once it was removed |
| Cash and colleagues, Trends in Cognitive Sciences (review) | 2026 | Review of existing studies | Offloading can erode specific skills, though core abilities such as working memory seem resilient |
The Turkish study holds the most useful lesson. A second version of the chatbot, designed to give hints rather than answers, largely avoided the learning loss. AI that does the work for you can weaken you, while AI that coaches you need not.
The other side
None of this means AI is making us stupid. The same research shows real gains when people use it well.
A 2025 study led by MIT's Jackson Lu, reported by the American Psychological Association, looked at 250 consulting employees in China. Those given ChatGPT produced more creative work, especially those with strong metacognition, the very skill Rock recommends.
The Trends in Cognitive Sciences review also cautions against panic. As reported by PsyPost, its authors judge a fundamental loss of intelligence unlikely. What fades is the specific skills we stop practising, and how AI is designed and used makes the difference.
The evidence has clear limits too. The MIT brain study is a preprint, not yet peer reviewed, and only 18 people completed its final session. The Microsoft survey relied on self-reports and cannot prove cause and effect.
Rock's 5 to 10 per cent figure is offered as a rough estimate, and the UN piece cites no survey behind it. Treat it as an expert's warning, not a measurement.
How to keep your brain in the game
These are practical habits, not research findings, though several echo the studies above.
- Try first, then ask. Write your own formula, SQL query or outline before you open the chatbot. Michael Gerlich of the Swiss Business School, quoted by the APA, recommends this order, then asking AI for feedback on your attempt.
- Ask for hints, not answers. Tell the chatbot not to solve the problem but to nudge you towards the solution. Better still, ask it to quiz you on the topic until you get it right.
- Verify before you trust. Check AI figures against your source data and recalculate a few by hand. Open every reference it gives you, because checking is now part of the job.
- Choose what to hand over. Let AI format tables, tidy meeting notes and draft routine emails. Keep the analysis, the judgement and the final recommendation for yourself.
- Protect deep-work time. Block 60 to 90 minutes a day with notifications off and no summaries. You cannot think deeply while being interrupted every few minutes.
- Explain it back. After AI helps you, explain the answer in your own words or teach it to a colleague. If you cannot, you have not learnt it yet.
- Keep people close. Talk problems through with classmates, mentors and colleagues. Let a chatbot help you prepare for those conversations, not replace them.
Closing thought
AI is not going away, and nobody should wish it away. For Nigerian learners and professionals, it can be a patient tutor, a research assistant and a writing coach in one.
Rock compares this moment to the early days of cars, before traffic laws existed. The rules for AI are still being written, and the most important ones are personal.
So use AI to stretch your thinking, never to skip it. The goal is not to be faster than everyone else. It is to stay sharp enough to know when the machine is wrong.
Sources
- AI is changing the way we think. The question is whether we'll notice in time, UN News, 29 October 2026
- Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task, MIT Media Lab (arXiv preprint), 10 June 2025
- The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers, CHI 2025 (Microsoft Research and Carnegie Mellon University), April 2025
- Generative AI without guardrails can harm learning: Evidence from high school mathematics, PNAS (record on IDEAS/RePEc), July 2025
- Without Guardrails, Generative AI Can Harm Education, Knowledge at Wharton, 27 August 2024
- Is AI making us stupid through cognitive offloading? New review explores the evidence, PsyPost, 15 July 2026
- How AI is reshaping human skills and thinking, Monitor on Psychology (American Psychological Association), 1 July 2026
- How AI and Human Behaviors Shape Psychosocial Effects of Chatbot Use: A Longitudinal Randomized Controlled Study, MIT Media Lab and OpenAI (arXiv preprint), 21 March 2025
- Nigeria emerges global leader in AI adoption for learning, entrepreneurship — Google, BusinessDay, 21 January 2026