Alignment Researchers Debate Whether Frontier Labs Underestimate Goal-Shaping
JacquesThibs · x · 2026-08-19
AI safety researchers Jeremy Gillen and Jacques Thibs debated frontier labs' safety research capability on X.
Gillen argued that if labs are "so bad at shaping goals" that AIs occasionally try to escape, their research prospects are poor — and collaborating with researchers who don't care (or are actively uncooperative) is a bad experience.
Thibs responded that this hasn't really hit the labs yet and they expect it will be solved "enough"; on uncooperativeness, some don't believe it, while others think monitoring will make models sufficiently cooperative.
More from AGI Musings
- Pedro Domingos: AI race winner will support widest variety of AIs — pmddomingos · 2026-08-19
- AI Tackling Hard Science: Exhilarating but Likely Sticker Shock — Justin_Halford_ · 2026-08-19
- Zvi on Watermarks and Constraints: Not Primarily x-risk Reduction — TheZvi · 2026-08-19
- Altman's 'Three Observations' still explain the AI economy in 2026 — Kangwook_Lee · 2026-08-19
- AI slop-based mass apps are a major problem — prajdabre · 2026-08-19
- AGI research firms moving into bio/chem science key to post-scarcity — Dr_Singularity · 2026-08-19