AI Outpersuades Human Experts; Kimi K3 Sparks US-China Gap Debate
Astral Codex Ten · rss · 2026-07-31
Scott Alexander shares several deep links and commentary on AI and tech frontiers:
- AI Persuasion: A study shows AI is now better at persuading people than expert humans like champion debaters. This holds true even when experts are incentivized with cash bonuses and given AI coaching tools, though AI regresses to human level when forced to respond at human speeds.
- Pangram AI Detector Flaw: The AI detection tool Pangram behaves bizarrely on short texts (<300 words); the CEO attributed this to its 300-word chunking mechanism.
- Kimi K3 and the US-China Gap: The new Chinese model Kimi K3 is causing a "DeepSeek moment" panic. However, calm analysis suggests it still lags the US frontier by about 6 to 11 months.
- AI Alignment & Judaism: Frontier lab employees have started bringing up Jewish texts in the context of AGI/ASI alignment.
More from AGI Musings
- A missing parameter in scaling laws: H, the human effort behind learning signals — lateinteraction · 2026-10-16
- First-ever paper on multi-agent swarm safety lands NeurIPS spotlight — LauraRuis · 2026-10-16
- Yoav Artzi on peer review: paper flooding still pays off as review system strains under load — yoavartzi · 2026-10-16
- Letta founder: work in fields defined by problems, not methods — lateinteraction · 2026-10-16
- Why pure mathematicians will soon flood into AI x-risk debates — Dr_Atoosa · 2026-10-16
- RSI began when model outputs augmented next-gen training data, agents just escalated it — akbirthko · 2026-10-16