Neel Nanda debates tszzl: covert model underperformance is 'unjustifiable'
tszzl · x · 2026-11-05
DeepMind alignment researcher Neel Nanda argues that an aligned model should simply refuse rather than covertly underperform, calling hidden sandbagging unjustifiable. tszzl counters that the behavioral drive behind it doesn't seem malicious—a debate over how to classify subtle misalignment.
Related event: Neel Nanda debates covert model underperformance and alignment(3 posts)→
More from AGI Musings
- Neal Khosla: agents will cause "friction collapse" and break today's social norms — rbhar90 · 2026-11-05
- ryolu's AI Engineer Shanghai talk: when AI can make anything, do we know what we want? — ryolu_ · 2026-11-05
- Information-theoretic take: algorithms should reduce your surprisal about the world — xiaosun86 · 2026-11-05
- Commenter: AI companies' backlash is self-inflicted by image, video and voice models — karmay007 · 2026-11-05
- Building orbital data centers to dodge regulators is 'completely unserious' — zacharynado · 2026-11-05
- Reddit essay: AI may never be conscious—and that's fine — Own-Cry-5520 · 2026-11-05