✳flâneur — a map of the web's best reading
How will we update about scheming? — LessWrong
lesswrong.com · 16,383 words · saved by 1 readers
I mostly work on risks from scheming (that is, misaligned, power-seeking AIs that plot against their creators such as by faking alignment). Recently,…
x How will we update about scheming? — LessWrong Redwood Research Deceptive Alignment Outer Alignment AI Curated 2025 Top Fifty: 5 % 177 How will we update about scheming? by ryan_greenblatt 6th Jan 2025 AI Alignment Forum 44 min read 21 177 Ω 86 I mostly work on risks from scheming (that is, misaligned, power-seeking AIs that plot against their creators such as by faking alignment). Recently, I (and co-authors) released " Alignment Faking in Large Language Models ", which provides empirical evidence for some components of the scheming threat model. One question that's really important is how
Explore this link on the map →related reading
- How will we update about scheming? - by Ryan Greenblattredwoodresearch.substack.com
- Thinking about reasoning models made me less worried about scheming — LessWronglesswrong.com
- Scheming AIs Will AIs fake alignment during training in order to get power?arxiv.org
- Reducing risk from scheming by studying trained-in scheming behavior — LessWronglesswrong.com
- Thinking about reasoning models made me less worried about scheming — LessWronglesswrong.com
- Frontier Models are Capable of In-context Scheming — AI Alignment Forumalignmentforum.org
- Catching AIs red-handed — LessWronglesswrong.com
- AI 2027ai-2027.com
- Current AIs seem pretty misaligned to me — LessWronglesswrong.com
- Training-time schemers vs behavioral schemers — LessWronglesswrong.com
- Current AIs seem pretty misaligned to meblog.redwoodresearch.org
- The case for ensuring that powerful AIs are controlled — LessWronglesswrong.com