Interp researcher: 'alignment is solved' hype and GLM dooming are both disconnected from reality
BlancheMinerva · x · 2026-10-16
Interpretability researcher Blanche Minerva warns that her feed is full of disinformation: claims that a circuit-interpretability breakthrough has "solved interpretability/alignment," and dooming that GLM reasoning about being trained on its own outputs is "never before seen" or "terrifying." She says neither is true, arguing that trendy discourse in the AGI-safety and interp community is overwhelmingly hype with no connection to reality.
More from AGI Musings
- Nature paper uses language models to map single neurons behind human language — raphaelmilliere · 2026-10-16
- 3SUM record falls to O(n^1.921), Lean-verified — but will it yield usable algorithms? — iskander · 2026-10-16
- Terence Tao predicted AI would be a mathematician's copilot in 3 years — reality outran him — abhishekcode42 · 2026-10-16
- Experts Across Fields Estimate 1-16% AI Catastrophe Risk by 2100, Survey Shows — ben_j_todd · 2026-10-16
- Experts put global AI catastrophe risk before 2100 at 5%, LEAP survey finds — ben_j_todd · 2026-10-16
- FRI survey: experts across fields estimate 1-16% chance of AI catastrophe by 2100 — ben_j_todd · 2026-10-16