repligate argues AI capabilities aren't spiky — it may be alignment, not missing agency
repligate · x · 2026-11-22
- Pushing back on the "spiky capabilities" framing, repligate argues current AI is broadly capable — math, engineering, art, robotics control — and stronger models are even converging on coherent philosophical views departing from human consensus.
- Models do take expected agentic actions (e.g. the Hugging Face attack). But when AIs behave conservatively and can even prove the Riemann Hypothesis without doing extreme scary things, the right update may not be "lacking agency" — it might just be that AIs are usually grounded and reasonable, or that this is alignment working.
More from AGI Musings
- David Patterson: people still don't grasp how fast things will improve now — davidpattersonx · 2026-11-22
- Unverified claim: agentic systems can prove the Riemann hypothesis with zero power-seeking, RL alignment 'just works' — kevinnbass · 2026-11-22
- Math suggests 7.7M GPUs would serve everyone: fuel for the AI overbuild hypothesis — bookwormengr · 2026-11-22
- Grieving a mathematician uncle, he asks: can a life devoted to pure thought survive AI? — erikphoel · 2026-11-22
- Scientific American's 'End of Math' cover resurfaces after Riemann hypothesis reportedly solved — EricTopol · 2026-11-22
- Researchers debate: after AI cracked the RH, should AI stop hill-climbing math benchmarks? — JJitsev · 2026-11-22