(By “most promising” I mostly mean “not obviously making noob mistakes”, with the central examples being “any Proper Noun research agenda associated with a specific person or org”.)
(By “formal” I mean “involving at least some math proofs, and not solely coding things”.)
Asking because the field is both relatively-small and also I’m not sure if any single person “gets” all of it anymore.
Example that made me ask this (not necessarily a central example): Nate Soares wrote this about John Wentworth’s work, but then Wentworth replied saying it was inaccurate about his current/overall priorities.
[Question] Does anyone’s full-time job include reading and understanding all the most-promising formal AI alignment work?
(By “most promising” I mostly mean “not obviously making noob mistakes”, with the central examples being “any Proper Noun research agenda associated with a specific person or org”.)
(By “formal” I mean “involving at least some math proofs, and not solely coding things”.)
Asking because the field is both relatively-small and also I’m not sure if any single person “gets” all of it anymore.
Example that made me ask this (not necessarily a central example): Nate Soares wrote this about John Wentworth’s work, but then Wentworth replied saying it was inaccurate about his current/overall priorities.