KNC started as a reading group on the intersection between computational neuroscience and machine learning. Since then, most of our members have transitioned to AI safety work. We’ve restarted this reading group with the primary aim of helping students and early-stage researchers:

• read through the existing introductory AI safety literature ( see here )
• keep up to date with new research
• work on smaller AI safety projects

We are open to anyone at any level of skill or interest in AI safety, and we primarily work on interpretability and mesa-optimization.

Please contact us at join (at) knc (dot) ai if you’re interested in joining a reading group, giving/receiving mentorship, or giving a talk and reciving feedback!