Event Details

Speaker: Herbert Chang

Abstract:
The 2024 USA Presidential Elections in the United States marked the first use of sophisticated generative AI to generate information and misinformation about candidates and their policies. In a series of projects, we combined computational and experimental efforts to comprehend the role of visual platforms in this election. We synthesize findings from interconnected studies that provide a comprehensive picture of visual content on actual policy preferences.

First, we characterize the use of synthetic media. Using a dataset of 239,526 Instagram images spanning seven months toward the election (04/05/2024 – 11/05/2024), we examine the impact of different content types on user engagement. We employ zero-shot labeling, deep learning, and OpenAI’s generative classifier to identify visual themes and synthetic content. We find that AI-generated content alone does not increase exposure and engagement, but when combined with memes, it generates a strong synergistic effect. We identify partisan asymmetries in generative use.

Second, drawing on ecologically-valid visuals from the previous studies, we conducted a large-scale survey experiment. Based on conjoint analysis, we combine a framing experiment with randomly assigned text and images to examine the causal effects of AI-generated content on actual policy preference. We conclude by discussing the methodological merits of combining computational social science and LLMs with contemporary survey techniques, especially for producing rapid, ecologically-valid research during critical events.

Virtual Link

Webinar Passcode: 245670

Webinar ID: 918 8531 1055

Speaker Bio

Herbert Chang is an Assistant Professor of Quantitative Social Science, Computer Science, and Mathematics at Dartmouth College, and a Forbes Under 30 Honoree in Science. His research studies how emerging technologies impact democratic behavior. He has published more than 35 peer-reviewed articles on misinformation, the offshore financial networks, and the civic impact of AI. His work has been featured in the New York Times, Washington Post, and Scientific American.

Host: David Chu

POC: Peter Zamar