Critics Slam AI Agent Science Claims: No Reputational Cost for Wasted Compute
ludwigABAP · x · 2026-10-06
The context: ValsAI reported that 90+ Opus 5.5 agents uncovered two room-temperature magnetic semiconductor candidates in 3 days of simulations, including KV[Cr(CN)₆], which was synthesized back in 1999 with its spin-sorting ability "hiding in plain sight for 27 years."\n\nThe retweeted critique from doomslide is blunt: one problem with replacing academic scientists with AI agents is the lack of reputational damage — you can waste days of cloud capacity producing nothing and still get rewarded by clueless investors. You don't even need an expert to see through it; just ask Chat.
More from AGI Musings
- Losing your job to a robot may free the life the job was stealing — rand_longevity · 2026-10-06
- Investor warns: don't start a personal agents startup without distribution or tier-1 capital access — vaibhavbetter · 2026-10-06
- UK essay argues AI acceleration, not regulation, is safest path for Britain's global leverage — dc_lawrence · 2026-10-06
- Synthetic data is the biggest opportunity in AI coding, argues long thread — creatoroff · 2026-10-06
- Video essay imagines interstellar minds traveling as light, in a Feynman-Moravec-Deutsch collaboration — anderssandberg · 2026-10-06
- Only Math and CS People Talk About 'Solving Alignment,' Argues Dan Faggella — danfaggella · 2026-10-06