Garrison Lovely: the alignment problem was a mistake—governance is the real fix
GarrisonLovely · x · 2026-10-24
Garrison Lovely argues RLHF was built at OpenAI as a safety technique, yet it also made LLMs useful enough to produce ChatGPT. His case: solving alignment makes the product better and the weapon stronger, so it is "neither necessary nor sufficient"—governance, not alignment research, is what could actually address the risk.
More from AGI Musings
- The most consistent criticism of AI companies: if you believe in catastrophic risk, why build so fast? — GarrisonLovely · 2026-10-24
- Kenneth Stanley defines open-endedness: processes that innovate indefinitely, like evolution — GregKamradt · 2026-10-24
- Gary Marcus amplifies AI bubble critique as hedge funds log widest daily drawdown since Covid — GaryMarcus · 2026-10-24
- OpenAI's 2019 prediction that complex intelligent agents would emerge gives netizens goosebumps — flowersslop · 2026-10-24
- Mathematician Danny Calegari: LLMs Mirror You Back, But Behave Better — fortnow · 2026-10-24
- No One Is Impressed by Good Results: Why AI-Proved Theorems Teach You Nothing — fortnow · 2026-10-24