"Thomas Kwa's MIRI research experience" by Thomas Kwa and others
LessWrong (Curated & Popular)
00:00
Deconfusion, Alignment, and Research Experience
This chapter explores the research methodology of deconfusion, particularly in the context of alignment problems in AI. The speaker discusses tensions between conflicting intuitions and the importance of concrete cases and theorems. They also reflect on the value of research experience and mentorship, as well as the challenges of turning conceptual problems into concrete ones.
Play episode from 21:43
Transcript
