Professor let students use LLMs in a zero-stakes oral exam — some still cheated
ipeirotis · x · 2026-10-04
NYU professor Ipeirotis had students discuss their already-graded submissions with a voice AI agent. The exam was pass/fail — just showing up was enough — and LLMs were allowed. Yet recordings revealed some students going silent 20–30 seconds per question, typing audibly, then reading out fluent answers. With zero incentive to cheat, he found it deeply discouraging, likening it to handing your phone to a friend to fake a Strava run. He pursued no charges but didn't expect to need restrictions for such a low-key exam.
Related event: NYU Professor Dismayed as Students Still Cheat in Zero-Stakes AI Oral Exam(2 posts)→
More from AGI Musings
- How Well Do We Really Understand Neural Networks? An AI Interpretability Debate — burny_tech · 2026-10-04
- Rationalists have saner norms, so don't trust other communities' rejects — ctjlewis · 2026-10-04
- Reward hacking has no general engineering fix — it's an old RL problem, not sloppy engineering — burny_tech · 2026-10-04
- AI tutors could break education's one-teacher limit, argues viral thread — r0ck3t23 · 2026-10-04
- Reddit post predicts model capability gaps could vanish within 1,000 days — BrennusSokol · 2026-10-04
- Anthropic study: robots cost-competitive for just 0.3% of US work hours — VraserX · 2026-10-04