Tom Dietterich: AI safety hinges on novel, ill-defined tasks AI can't optimize
tdietterich · x · 2026-09-09
In a debate sparked by Melanie Mitchell's "Deep Blue–Kasparov moment" remark, Tom Dietterich agrees AI will beat humans on well-defined tasks within a decade, but argues novel and ill-defined situations lack benchmarks and training data, making them hard to optimize for — and safe AI requires solving them too. The counterparty cites METR's time horizon data to argue the crossover will come faster.
More from AGI Musings
- Will AI Turn Mathematicians Into Shakespearean Scholars, Just Reinterpreting? — inductionheads · 2026-09-09
- As AI churns through conjectures, researcher warns academic math may already be effectively dead — secemp9 · 2026-09-09
- Fortnow: his 'Our Days Are Numbered' math-AI prediction is coming true fast — fortnow · 2026-09-09
- After Navier-Stokes: Researcher Points to Medicine and Biology as AI's Next Big Targets — finbarrtimbers · 2026-09-09
- Math Professor Roasts Anima Anandkumar's Navier-Stokes Reply-Guying as 'Crypto Scam'-Like — suchenzang · 2026-09-09
- When image/video generation is basically free, what stays scarce? Taste, identity and ownership — remixeconomy · 2026-09-09