Luke Hewitt

Luke Hewitt specializes in Bayesian machine learning, with a particular interest in public opinion and belief change. 

He is currently a visiting scholar at the Stanford Center on Philanthropy and Civil Society, collaborating with the Polarization and Social Change lab. In addition, he is the co-founder and director of Rhetorical, a research nonprofit dedicated to aiding public communication campaigns in running experiments to enhance the persuasive impact of their messaging. His role as co-PI for the SSRC Mercury Project team involves tackling health misinformation with community-crafted messaging.

His PhD, advised by Josh Tenenbaum of MIT Brain and Cognitive Sciences, was primarily focused on the development of scalable Bayesian methods for explainable AI. His thesis entailed conducting the largest RCT meta-analysis of political advertisements to date, in collaboration with David Broockman, Alex Coppock, and Ben Tappin.

Hewitt also contributed to public opinion methodology at Swayable, where he designed their 2020 national polling. Notably, his efforts led to the accurate prediction of Biden’s vote share to half a percentage point, compared to the 2pp bias of 538’s prediction.