AI Safety Support recently hosted a closed, introductory talk by Connor Leahy on Promising Paths to Alignment.
The talk covers: (i) why Alignment is such a difficult problem, (ii) current approaches to solving it, and (iii) some info on Connor’s new alignment research startup, Conjecture.
I highly recommend this talk for developing a better understanding of the technical alignment research landscape—particularly for those considering or pursuing a related career.
[Link post] Promising Paths to Alignment—Connor Leahy | Talk
Date of recording: Tuesday May 10, 2022
Description
AI Safety Support recently hosted a closed, introductory talk by Connor Leahy on Promising Paths to Alignment.
The talk covers: (i) why Alignment is such a difficult problem, (ii) current approaches to solving it, and (iii) some info on Connor’s new alignment research startup, Conjecture.
I highly recommend this talk for developing a better understanding of the technical alignment research landscape—particularly for those considering or pursuing a related career.