This abstract is part of a submission on a panel GOVAI, “Preferences over AI Regulation”.
As advancements in the capabilities of Artificial Intelligence (AI) continue, citizens around the world are demonstrating increasing levels of both concern but also hopes about how AI will affect their daily lives. A prominent policy response advanced by many is greater government regulation and scrutiny of the adoption of AI technology by governments and private actors. This paper presents new hypotheses and fresh evidence about the determinants and support for such regulation across a wide variety of AI domains (including data privacy, misinformation, security, labor substitution, and discrimination). We theorize that support for regulatory support should be shaped by three sets of arguments: about the risks that AI poses, about the costs of regulation, and about the non-government alternatives available to regulate AI, and that such arguments should carry similar weight across the relevant AI policy sphere in question. We gather new survey data from large samples in six countries where AI investment and discussion are increasing (China, Germany, India, Japan, the United Kingdom, United States) from April-August 2024 to test our hypotheses. We find strong support that all three types of arguments can affect AI regulatory support across many policy domains for most countries. However, across most policy domains, invoking the regulatory costs decreases regulatory support the most, indicating that citizen concern about AI can be reduced if key costs are made salient.