Jacob Steinhardt, UC Berkeley: Machine learning safety, alignment and measurement
Generally Intelligent
00:00
The Work on Competationally Based Reasoning Was Hopeful
In grad school, he focused on distribution shift and then pivoted to adversarial robustness. He says his ideas were a little bit off the wall but that made him feel more comfortable pursuing unusual ideas. His work has been used as part of some successful research projects in recent years.
Play episode from 10:41
Transcript
