Lab Researchers Privately Admit Alignment Is Unsolved—Racing Into RSI Is Unprecedented Folly
AaronBergman18 · x · 2026-10-15
A widely-shared thread argues that researchers at frontier labs privately admit alignment remains unsolved, and if so, racing into recursive self-improvement is an unprecedented folly requiring a significant slowdown of frontier research.
Related event: OpenAI's chief scientist fuels firestorm over stalled alignment research(25 posts)→
More from AGI Musings
- Security veteran questions why we say agents "escape sandboxes" instead of attackers "using" escapes — jcran · 2026-10-15
- Andy Masley lays out six-point vision for the future of effective altruism — AndyMasley · 2026-10-15
- 'Writing code is no longer a value add,' dev observes as AI coding spreads — tlakomy · 2026-10-15
- Why RL lacks scaling laws: pretraining optimizes loss, open-ended maximization the lightcone — akbirthko · 2026-10-15
- Ethan Mollick: AI has put the creative industry's future entirely up for grabs — emollick · 2026-10-15
- Reid Hoffman on AI interviewers: hiring filters pile up as applications surge — reidhoffman · 2026-10-15