AI isn’t making you smarter. It might be doing the opposite 🤖

Cliquez ici pour lire en français

It’s 11pm in Yaoundé. Brice, a computer science student, opens ChatGPT to finish a homework assignment he could have written himself. A few minutes later, the assignment is done — but Brice barely remembers what he just submitted. You may have lived through this scene yourself, or watched someone close to you go through it.

ChatGPT, Gemini, Copilot: these tools have taken over our daily lives faster than almost any technology before them. We were sold on the promise of productivity, of pushing our limits further than ever. But over the past year, researchers have been raising a less flattering alarm: the same tools might be quietly eroding our ability to think for ourselves. The phenomenon even has a name now — cognitive debt. And Norway, a pioneer of digital classrooms for two decades, just made a decision that says a lot about how seriously this is being taken.

What MIT’s study reveals about your brain 🧠

The conversation traces back to a study from MIT Media Lab, first published as a preprint in June 2025 and still widely discussed a year later. Led by researcher Nataliya Kosmyna, the team followed 54 participants split into three groups for an essay-writing task: one using ChatGPT, one using a regular search engine, and one relying on nothing but their own thinking. Over four months, their brain activity was tracked using EEG.

One finding stands out clearly: brain connectivity dropped as external assistance increased. The « brain-only » group showed the richest, most widespread neural networks, the search-engine group sat in between, and the ChatGPT group showed the weakest engagement overall. Participants who used AI also reported a lower sense of ownership over their own essays, and struggled more to accurately quote what they had just written themselves.

One detail stands out. In a fourth session, the researchers swapped the groups. Participants who moved from ChatGPT to « brain-only » still showed weaker neural engagement, as if the habit had left a trace. Those who moved the other way, from « brain-only » to ChatGPT, regained stronger memory performance.

A caveat is in order: the sample remains small — 54 participants, only 18 of whom completed the fourth session — and the study has not yet gone through independent peer review. The authors themselves acknowledge this and call for larger-scale research. What the study establishes with more confidence is a worrying correlation, not yet definitive proof of lasting damage.

Norway sounds an unprecedented alarm 🇳🇴

If MIT’s study fueled debate in academic circles, it was a political decision that pushed the issue into mainstream conversation. On June 19, 2026, Norwegian Prime Minister Jonas Gahr Støre announced a near-total ban on generative AI tools in schools for children aged 6 to 13, effective from the start of the new school year in late August 2026.

For teenagers aged 14 to 16, AI use remains possible, but only under a teacher’s supervision. From age 17 onward, students are instead encouraged to learn to use these tools independently and responsibly, in preparation for the job market.

The government justified the decision as necessary to protect foundational skills — reading, writing, arithmetic — before introducing tools that could let younger students bypass these essential steps. The move follows a path Norway had already set in motion: the country banned smartphones in schools back in 2024, a measure that, according to the government, reportedly reduced bullying, improved grades, and lowered the number of teenagers seeking help for psychological distress — with the strongest effects reportedly seen among girls. Oslo is now preparing a ban on social media for anyone under 16, modeled closely on Australia’s approach.

What this means for Yaoundé and Douala 🌍

Norway is obviously not Cameroon. School-wide digital infrastructure has been the norm there for years — a reality far removed from ours, where access itself is still the main challenge. But the question raised by this decision concerns us directly, perhaps even more so: if a country this digitally advanced feels the need to slow down AI use among its youngest students, what does that say about a context where this same use is spreading with no framework at all, neither at school nor at home?

More and more students in Cameroon are turning to ChatGPT for their homework, using a smartphone and a bit of mobile data, often without telling a teacher or a parent. The motivation is entirely understandable — saving time, securing a grade, avoiding the blank page — but it’s becoming a habit without any collective reflection on its long-term effects, partly because no study has yet looked specifically at our context. This isn’t a call for a Norwegian-style ban, which wouldn’t make sense here. It’s a call for some vigilance, from families and schools alike, before the habit fully sets in unchecked.

Taking back control without giving up AI 🛠️

Cognitive debt isn’t inevitable, and it’s not a reason to avoid these tools altogether. It’s a matter of how you use them. A few simple habits can make a real difference.

First, write a rough first draft yourself before turning to AI, even an imperfect one. That initial effort is what engages your thinking; AI can then step in to refine, restructure, or enrich it — not replace the starting point.

Second, treat AI as a sparring partner that challenges your ideas, rather than a ghostwriter that replaces them. Asking « what’s weak about my reasoning here? » produces a very different outcome than asking it to « write this for me. »

Third, make a habit of putting whatever AI gives you back into your own words before using it, submitting it, or sharing it. If you can’t explain something without rereading it, that’s a sign you haven’t really absorbed it — and that’s exactly what cognitive debt measures.

Tech should make us more capable, not more dependent 🦾

What’s really at stake behind cognitive debt isn’t a case against artificial intelligence. It’s a reminder: a tool remains a tool, and it’s how we use it that decides whether it makes us more capable or more dependent. Norway chose to settle the question by law, for its youngest students. We don’t have the same framework, or the same educational urgency — but we carry the same individual responsibility: deciding, consciously, when to think for ourselves and when to delegate. Tech remains our entry point. But staying in control of our own thinking is what should remain our destination.

💬 Where do you draw the line with AI? Do you use it to think with you, or to think for you? Whether it’s for school, work, or your kids — tell us in the comments how you navigate that balance.