IEEE Members: Free
Non-members: FreeLength: 00:57:31
05 Jun 2023
Peter Norvig, Stanford University Human-Centered AI, USA; ABSTRACT: Educators have long known that tutoring that is adapted to each individual student’s pace and interests is superior to a lecture that tries to reach every student at once. But we have not had the resources to give every student a qualified tutor. In the past few months, we have seen tantalizing examples of large language models that might someday be able to fill this role. This talk examines how close they are, and what remains to be done.