tszzl: no fundamental alignment progress since RLHF, capabilities race ahead
tszzl · x · 2026-10-13
tszzl argues there have been no fundamental advancements in alignment since RLHF — most of the field remains folk theory rather than a scientific or engineering discipline, and the prediction that capabilities would race far ahead of interpretability and alignment has come true.
Related event: tszzl: Alignment Has Stalled Since RLHF, Frontiers Should Slow Down(4 posts)→
More from AGI Musings
- Language is a latent space for world knowledge, at the cost of information loss — docmilanfar · 2026-10-13
- Thirty years of moving the goalposts: what AI keeps revealing about humanity — r0ck3t23 · 2026-10-13
- When you dismiss a working method as 'naive,' you may be missing something non-obvious — tak3sh8 · 2026-10-13
- The day we can't tell cricket sounds are real: AI and the trust divide — spilltrend · 2026-10-13
- Describe a Machine in One Line, Get a 3D Model You Can Pull Apart — alvelda · 2026-10-13
- Will DePue: all progress will soon be data bottlenecked, not intelligence bottlenecked — willdepue · 2026-10-13