Persuasion at Scale: Machine Learning, Causality, and the Information Ecosystem (Lessons Learned)

What happens when you bring causal inference, machine learning, and AI together with political science and media studies in a single classroom? This talk shares lessons learned from "Persuasion at Scale," a new course developed by Eunji Kim and Chris Wiggins that integrates quantitative methods from both academic research and industry practice to understand persuasion in the modern information environment.

The curriculum traces a five-part journey: from the universal challenge of quantifying messy human behavior into analyzable variables, through experimental methods (RCTs and A/B testing) and observational causal inference (IV, DiD, RDD), to optimization techniques used in industry (bandits, recommendation systems), and finally to the emerging frontier of AI-generated persuasion. Applications span political science, communication, journalism, marketing, and public health, with hands-on work in R and Python via Google Colab.

Three insights emerged. First, academic researchers and industry practitioners increasingly face identical methodological challenges with the same mathematical foundations. Second, students can handle contested topics and methodological uncertainty if instructors are honest about both the power and limits of rigorous methods. Third, a new field may be emerging at the intersection of these disciplines, one focused on the mathematics of both interpreting the world and changing it. The talk offers reflections relevant to both curricular innovation and research collaboration across the academic-industry divide.