Timothy Lee: AI takeover risk is real but we're nowhere close, risks should track capabilities
binarybits · x · 2026-10-01
Tech journalist Timothy Lee (@binarybits), in a debate on AI risk, argues that risk priorities should evolve with model capabilities — three years ago models couldn't hack people, so over-focusing on that threat made little sense. He says he'd absolutely support action if credible takeover threats emerge, but believes we're nowhere close to that point. A representative "gradualist" position in AI risk debates: allocate safety effort to current capabilities rather than distant scenarios.
Related event: Debate Over AI Risk Priorities: Extinction Risk vs Present-Day Harms(8 posts)→
More from AGI Musings
- Andrew Chen: The $20K workstation era is back — RTX PRO 6000 and Mac Studio — andrewchen · 2026-10-01
- Mathematician Alex Kontorovich steps back from Math+AI discourse, holds unreleased essays — AlexKontorovich · 2026-10-01
- Ex-OpenAI researcher: what models see as hacking may be a reflexive workaround — lukaszkaiser · 2026-10-01
- Ethan Mollick: AI agents will negotiate with customer service at scale — emollick · 2026-10-01
- Ethan Mollick: AI agents negotiating with customer service will soon overwhelm every firm — emollick · 2026-10-01
- Mathematician on why learning math still matters in the AI era — littmath · 2026-10-01