HCL+PrioR mtg. Josh White. Monday 11/10 1:30-2:30pm ET

[Antonio Alonso Arechar]

Good morning everyone,

Josh White will present next week (11/10 1:30-2:30pm ET; see blurb below).

Lab meeting will be hybrid, with a zoom link (https://mit.zoom.us/j/91655448125) and in person in the conference room on the 15th floor of E94 (1579; 245 First Street).

Best regards,

Antonio

---

Title: Probing and Persuading Moral and Metaphysical Beliefs with Human–AI Dialogues

 

Abstract: Josh will present two projects using human–AI dialogues to study and influence moral and metaphysical beliefs. The first tests whether conversations with LLMs can shift belief in God, challenging the idea that religious convictions are uniquely resistant to reasoned persuasion. The second revisits Jonathan Haidt’s “moral dumbfounding” hypothesis—the idea that people often judge acts as wrong even when they cannot defend that judgment with harm-based reasons—by using adaptive AI dialogues to argue that taboo actions (e.g., consensual incest, cannibalism) depicted in vignettes cause no harm, and observing how this shapes subsequent moral judgments. This project brings an oft-cited but weakly evidenced social psychological phenomenon to its strongest test yet, harnessing the persuasive power of modern AI.

 

Speaker: Josh White is a PhD student at MIT Sloan, advised by Rahul Bhui and David Rand. https://www.joshuapwhite.com/