Machine Intelligence Research Institute

Research program activities - we spent this year continuing our existing Research agenda and expanding work on new Research directions. Research staff were particularly active in posting technical writeups on our ai alignment forum. Our progress on all our Research directions and the supporting infrastructure was steady and significant, in line with our expectations. Significant Research publications include "delegative reinforcement learning: learning to avoid traps with a little help and "risks from learned optimization in advanced Machine learning systems"

Outreach - our outreach activities this year were focused on running workshops that exposed and oriented technically minded people to ai risk and various potential paths for Research. Workshops we hosted included: ai risk for computer scientists, a four-day workshop series designed to introduce technically minded people to thinking tools for reasoning about ai risk; miri summer fellows, a two-week retreat to bring people up to speed on problems related to ai alignment, train research-relevant skills, and investigate open problems in the field. Outreach publications include "alignment Research field guide and "ai safety reading list". We attended the global governance of ai roundtable event.