80,000 Hours' Wiblin urges funding grassroots AI risk groups, sparking 'epistemics' drama
zetalyrae · x · 2026-11-14
Robert Wiblin of 80,000 Hours published a long thread urging donors to fund grassroots AI risk advocacy groups (e.g., Humans in Control, PauseAI UK) — not his first-choice method, but by far the most underfunded part of the AI risk ecosystem, since major donors shy away while large shares of US/UK publics already support pacing AI.
He also admits the risks (see screenshot): grassroots movements tend to have low epistemic hygiene, audience capture, rage/purity spirals, and will inevitably advocate some ineffective or harmful policies.
Retweeter GWHayduke97 pushes back hard, calling this the strongest example of 'good epistemics' being rat-speak for 'agrees with me' — a snarl term to marginalize opponents of what he views as misanthropic worldviews. The fight exposes a deep strategic and discursive split within the AI safety community.
More from AGI Musings
- 10 Years of Postgres Expertise: 'Claude Just Made the Whole Team a Postgres Expert' — RealGeneKim · 2026-11-14
- Guardrails redefined: agent-written tests are eroding the term's meaning — blaizedsouza · 2026-11-14
- Speculation: what would a real nation-state AI-assisted cyberattack look like today? — teortaxesTex · 2026-11-14
- 'AI' conflates memorization with intelligence — a pointed one-liner — JosephJacks_ · 2026-11-14
- "We are living in a science fiction world": AIX risk debate over light-cone arguments — teortaxesTex · 2026-11-14
- Debate over a PhD student's AI safety thesis trouble: department or student at fault? — avt_im · 2026-11-14