Interpretability researcher amplifies sarcastic list of everything 'EAs got right'
NeelNanda5 · x · 2026-10-05
Neel Nanda retweeted Laneless's sarcastic rant: aside from pandemics, cryptocurrency, scaling laws, power seeking, instrumental convergence, CoT monitorability, mechanistic interpretability, RLHF, sycophancy, RL-induced reward hacking, emergent misalignment, and the insights that let a small offshoot of OpenAI employees overtake OpenAI — when have the EAs ever been right about anything?
A dense, ironic catalogue of the EA/AI-safety community's biggest wins, resonating in the alignment crowd.
More from AGI Musings
- Perspective warns AI can breed illusions of understanding and scientific monocultures — MichaelRetchin · 2026-10-05
- Kai-Fu Lee: After 100+ CEO interviews, the biggest AI transformation mistake is treating AI as another tech cycle — kaifulee · 2026-10-05
- Blogger calls out Anthropic's contradiction: tool or $2T-enslaved sentient being? — alexeyguzey · 2026-10-05
- DearCai CEO on 60 Minutes: AI labor shift will be the most consequential of our lifetimes — clarashih · 2026-10-05
- OpenAI DevDay's overlooked big idea: agents that take full responsibility like employees — jxnlco · 2026-10-05
- Musk calls humanity a "biological bootloader"; reply goes viral: I don't want to be one — davidmanheim · 2026-10-05