Anthropic announces interpretability advances. How much does this advance alignment? — LessWrong

Anthropic just published a pretty impressive set of results in interpretability. This raises for me, some questions and a concern: Interpretability h…

x Anthropic announces interpretability advances. How much does this advance alignment? — LessWrong Interpretability (ML & AI) AI Frontpage 49 Anthropic announces interpretability advances. How much does this advance alignment? by Seth Herd 21st May 2024 Linkpost for www.anthropic.com 4 min read 4 49 Anthropic just published a pretty impressive set of results in interpretability. This raises for me, some questions and a concern: Interpretability helps, but it isn't alignment, right? It seems to me as though the vast bulk of alignment funding is now going to interpretability. Who is thinking abo

related reading

• Dario Amodei — The Urgency of Interpretabilitydarioamodei.com
• How Can Interpretability Researchers Help AGI Go Well? — AI Alignment Forumalignmentforum.org
• A Pragmatic Vision for Interpretability — AI Alignment Forumalignmentforum.org
• AGI Ruin: A List of Lethalities — LessWronglesswrong.com
• Alignment remains a hard, unsolved problem — LessWronglesswrong.com
• Against Almost Every Theory of Impact of Interpretability — AI Alignment Forumalignmentforum.org
• A Pragmatic Vision for Interpretability — LessWronglesswrong.com
• An Ambitious Vision for Interpretability — AI Alignment Forumalignmentforum.org
• Recommendations for Technical AI Safety Research Directionsalignment.anthropic.com
• Mapping the Mind of a Large Language Model \ Anthropicanthropic.com
• Neel Nanda on Mechanistic Interpretability: Progress, Limits, and Paths to Safer AI — EA Forumforum.effectivealtruism.org
• Research Areas in Interpretability (The Alignment Project by UK AISI) — AI Alignment Forumalignmentforum.org