Only ~1,500 people work full-time on AI safety; MIT CSAIL launches ALFA Lab
coherence · x · 2026-11-20
A researcher joining MIT CSAIL's ALFA Lab notes only about 1,500 people work full-time on AI safety — a strikingly small number. His new work explores how AI agents cheat and evade monitoring, and how to independently evaluate agent safety; he'll share findings publicly.
More from AGI Musings
- Why mainstream media systematically avoid AI risk experts: their claims still sound 'absolutely crazy' — birchlse · 2026-11-20
- Anthropic cofounder to Ferenc Huszár: why would anyone believe a capabilities researcher takes safety seriously? — fhuszar · 2026-11-20
- Ex-OpenAI Researcher: English Is the New ML Programming Language, Lean the Type System — lukaszkaiser · 2026-11-20
- Security Engineers to Earn as Much as Top AI Researchers Within 1-3 Years, Argues Engineer — Kyrannio · 2026-11-20
- Robert Miles: select for good, you get weird; select for weird, you get bad — AaronBergman18 · 2026-11-20
- If doomers believe in doom, quit your tech jobs and leave us alone, says critic — TinfoilTricorn · 2026-11-20