How Media Coverage and Elite Communication Shape Public Opinion on AI Regulation

• In Event: Reimagining Democracy: Politics and New Technologies

Abstract

Attention to artificial intelligence (AI) is soaring around the world. The technology has the potential to profoundly impact many aspects of social and political life, including healthcare, education, transportation, employment, and finance. For the average news consumer, it is almost impossible not to come across media coverage of AI. Countless news articles have been published on artificial intelligence between 2022 and 2023, some highlighting the incredible promise of AI and others calling out its potential dangers and risks. In this study, we present a systematic analysis of U.S. media coverage of the development of artificial intelligence, as well as how elites, including elected representatives, cue the public about AI. We then examine how media coverage influences public opinion on AI regulation by presenting an original, pre-registered experiment embedded in a representative US survey. The experiment randomly manipulates both the content of media messages (positive/negative implications of AI/placebo) and the source delivering the messages (tech leaders, scientists, or elected officials), while focusing on three prominent themes covered by the media in the past year: racial bias in decision-making, automation in the labor market, and social effects of large language models such as ChatGPT. Using a set of attitudinal and quasi-behavioral measures, we assess the impact of these treatments on public opinion and support for policies regulating AI. Our research sheds light on the potential politicization of the public discourse on AI and offers insights into the political feasibility of the regulatory measures currently under debate.

Authors

• Shir Raviv, Columbia University

• Tamar Mitts, Columbia University