Research Workshops AI Workshop

Faculty and invited speakers present current AI and machine learning research. Meetings are open to all interested faculty and PhD students. The workshop will be held in-person. When possible, links to the workshop papers are posted to this page for printing. Occasionally, speakers opt not to circulate their papers, in which case the link will be unavailable. Workshop will take place on Wednesdays from 10-11am (unless otherwise noted), room TBD. Sign-up to our seminar listserv for weekly updates.

Please contact Lauren Doan (lauren.doan@chicagobooth.edu) with any questions.

Fall 2026

Date	Time	Room	Topic	Speaker	Institution
October 14, 2026	10-11am	219	Can Multi-Agent Systems Simulate the Complexity of Human Cognition and Social Behavior? Abstract & readings ↓	Rada Mihalcea	University of Michigan
October 19, 2026	10:30-11:30am	104	TBD	Eric Horvitz	Microsoft
October 21, 2026	10-11am	219	TBD	John Horton	MIT
October 28, 2026	10-11am	219	Unlocking the Knowledge in Deep Learning Models Abstract & readings ↓	Emma Pierson	UC Berkeley
November 4, 2026	10-11am	C25	TBD	Osbert Bastani	University of Pennsylvania
November 11, 2026	10-11am	C25	TBD	Nicholas Caputo	Johns Hopkins University
November 18, 2026	10-11am	C25	TBD	Nicholas Tomlin	TTIC

Talk Details

Oct 14 · Rada Mihalcea — Can Multi-Agent Systems Simulate the Complexity of Human Cognition and Social Behavior?

Abstract
Recent advancements in multi-agent systems have sparked a rapidly growing research area focused on simulating increasingly complex human behaviors — group consensus, implicit bias, and, in some cases, even cooperation or conflict. Yet, these systems are computational and artificial, entirely lacking the intrinsic consciousness, emotions, and social intuition that define human individuals and societies. In this talk, I will explore the evolving relationship between AI agents and human behavior, drawing on large-scale generative agent experiments, studies on bias in multi-agent interactions, and insights into misinformation and group behavior. I will also discuss the broader implications of these systems — not only for the future of AI but also for human-centered disciplines such as psychology, sociology, and business and economy, where they can challenge or facilitate our understanding of intelligence, agency, and collective decision-making.

Suggested Reading

Towards Implicit Bias Detection and Mitigation in Multi-Agent LLM Interactions (Findings of EMNLP 2024)
Mind the (Belief) Gap: Group Identity in the World of LLMs (Findings of ACL 2025)
Cooperate or Collapse: Emergence of Sustainable Cooperation in a Society of LLM Agents (NeurIPS 2024)

Oct 28 · Emma Pierson — Unlocking the Knowledge in Deep Learning Models

Abstract
Deep learning models often achieve strong predictive performance for reasons that remain opaque. This opacity both hampers scientific discovery — whose goal is not merely prediction, but understanding — and poses safety risks when these models are used in high-stakes domains. Here, we present an approach to interpreting the representations of deep learning models using sparse autoencoders and AI agents. We then discuss the applications of this approach both to improve model safety, and to accelerate discovery in the health and social sciences.

Suggested Reading

Sparse Autoencoders for Hypothesis Generation (ICML 2025)

Spring 2027

Date	Time	Room	Topic	Speaker	Institution
Speakers to be announced. Check back soon.