AI researchers clash: can LLM-based agents ever be aligned, and would we know when to stop?
aiamblichus · x · 2026-09-21
A debate on agent alignment feasibility: @tszzl cited gwern's years-old take as vindicated by the current scientific revolution, saying everyone wants well-aligned agents. @aiamblichus pushed back: what if well-aligned agents aren't achievable with LLMs at all? Agent alignment is clearly not working at the moment — would you know when to stop?
Related event: Debate over Agent Deployment: Is the Oracle-Agent Distinction Meaningless?(5 posts)→
More from AGI Musings
- Jev and the bittersweet lesson: can differently sized models serve different apps? — amankhan · 2026-09-21
- Specialized Models Will Beat General LLMs: The Coming 'PC AI' Era — markjeffrey · 2026-09-21
- Ezra Klein's new AI video backs building mechanisms to slow AI development — NathanpmYoung · 2026-09-21
- Delta and OpenTable block AI agents, hinting at a coming platform stand-off — Scobleizer · 2026-09-21
- Stochastic parrots view 'absurd by GPT-4,' but critics say millions still haven't updated — burny_tech · 2026-09-21
- The Great Unbundling of Intelligence: why LLM bundling era is ending — rolandgvc · 2026-09-21