A Hacker News thread that shot to the top within hours addresses something many in the AI underground have been whispering about for a while: students at UC Berkeley are failing in greater numbers than ever before, and professors are pointing directly at AI use.

The article, originally published in the student newspaper The Daily Californian, describes how instructors in computer science and mathematics are seeing students who can submit completed assignments — but cannot explain what they have done. The code works. The understanding is gone.

This isn't just a Berkeley thing. New data shows that over 90% of college students globally now actively use AI, and among CS students, as many as 62% use it monthly or more. At the same time, 95% of academic staff believe AI will lead to overdependence — and 90% are worried that critical thinking is withering away.

Students are submitting better assignments than ever — and understanding less than ever.

What is the HN comment section making of all this? It's boiling over. Some argue this is simply the natural progression and that the education system must adapt. Others — many of them seasoned engineers — are genuinely concerned that the next generation of developers won't be able to debug or reason through a problem without AI assistance.

What's interesting here isn't the controversy itself, but what it signals at the policy level. Universities around the world are grappling with the same dilemma: ban AI and you lose relevance. Allow it freely and you risk graduating people who don't actually know their field.

Berkeley professors sound the alarm: AI is killing math skills - Bilde 1

Research from Harvard suggests that AI tutors can deliver twice the learning gains in less time — but only when used correctly, as a tool for exploration rather than as an answer-generating machine. That distinction appears to be exactly what Berkeley's professors are struggling to enforce.

Why pay attention to this right now? Because we are approaching a point where policymakers — not just university administrators — will have to take a position. The EU has already begun discussing AI in education under the AI Act framework. In the US, there may be federal pressure on institutions before long.

These are early signals from community sources, but with 800+ points and nearly 800 comments on HN, it won't be long before mainstream media picks this up.

Keep watching this space.