Researcher warns AI could automate AI R&D within years while alignment lags
zetalyrae · x · 2026-11-17
A researcher argues AIs are uncomfortably close to automating AI R&D itself, likely within a few years, while still lacking robust alignment and control. He also notes the tension that near-term AIs may already be useful enough to substantially accelerate AI safety research — using AI to solve the alignment problem AI creates.
More from AGI Musings
- What status do mathematicians lose to AI proofs? 'Tao's grief is real' — georgejrjrjr · 2026-11-17
- Kai-Fu Lee: US and Chinese companies differ vastly in pursuing AI efficiency — kaifulee · 2026-11-17
- Coding Agents May Deliver 100-1000% Productivity Shock to Software Next Year — moultano · 2026-11-17
- AI safety interviews top 50M combined views; host teases Kokotajlo crossover — romanyam · 2026-11-17
- Ex-OpenAI's David Robinson: inside AI labs it's 'frenetic, people running on fumes' — connoraxiotes · 2026-11-17
- The case for games on personal agents: every platform ends up full of games — agentaya-ltd · 2026-11-17