Cognitive Science Talk Series: Language models as instruments of linguistic discovery with Christopher Potts, professor of linguistics, Stanford University

The AI story to date has primarily been about business and technology, but we are beginning to see that it can also be a story of scientific discovery, and even scientific revolution.

In this talk, I will offer a glimpse of this exciting possibility for the field of linguistics. My talk will focus on two related case studies: reformulating the argument from the poverty of the stimulus in light of what we now know about data-driven learning, and probing the distinction between possible and impossible human languages.

In both case studies, large language models function as powerful investigative tools, serving to refine the terms of the debate and provide novel sources of evidence.

Date: Tuesday, Sept. 22
Time: 12:30 p.m. - 1:50 p.m.
Location: Zoom
Meeting ID: 895 0983 3913

This talk is supported by the Broadening Participation in Cognitive Science Initiative Grant from the Cognitive Science Society.