Are teenagers getting weak in studies? Nithin Kamath flags worrying trend as AI enters classroom

Teenagers across much of the developed world are falling behind in reading and mathematics, according to OECD data highlighted by Zerodha founder and CEO Nithin Kamath. The decline, he says, raises a bigger question: could AI make people more dependent on machines for skills they once developed themselves?

The OECD's Programme for International Student Assessment (PISA) measures reading, mathematics and science skills among 15-year-olds. The chart shared by Kamath shows a marked decline in average scores across 23 OECD countries, particularly in reading and mathematics. The OECD estimates that 20 points on the PISA scale are roughly equivalent to one year of schooling.

Is COVID the whole story?  

Kamath acknowledged that the pandemic likely contributed to the deterioration, given prolonged school closures and disruptions to classroom learning. But he believes there may be a longer-term trend at work. "And now there's AI," he wrote.

His concern comes partly from personal experience. Kamath said he now runs much of his writing through ChatGPT, Claude and other AI tools because they can fix grammar and typos. That, he said, has reduced his incentive to get those things right himself.

"Over time, I genuinely feel like my writing without AI is getting worse," he said.

The AI learning dilemma  

Kamath said he has also asked Zerodha's CTO whether a similar pattern is emerging among the company's technology teams, with AI increasingly involved in coding.

The issue highlights a central dilemma for schools and workplaces: AI can make people more productive while potentially reducing the amount of cognitive work they do themselves.

For children, the trade-off is even more complicated. Students need to develop thinking, writing and problem-solving skills by struggling with problems and making mistakes. But they also need to learn how to use AI effectively because it will be an important part of their future education and careers.

The question, therefore, may not be whether children should use AI, but when and how they should use it.

The easy path  

Kamath argues that human behaviour could make the problem harder. "Given a choice between an easy thing and a hard thing, we'll almost always choose the easy path," he wrote.

AI makes that easy path almost frictionless. Students can generate answers, essays and code in seconds, while adults can use AI to draft emails, reports and presentations.

The immediate gains are clear. The long-term impact on skills is less certain.

Kamath ended his post by invoking Idiocracy, the satirical film in which human intelligence has declined so dramatically that an ordinary man from the past becomes the smartest person in the future.

His comparison is deliberately exaggerated, but the underlying concern is serious. If AI increasingly does the thinking for us, how do we ensure humans still learn how to think for themselves?

The PISA decline does not establish that AI is causing weaker academic performance. But as AI becomes embedded in education and work, the debate is shifting from how much AI people should use to how much thinking they should retain for themselves.