The Influence of Information and Audits on the Public’s Perception of AI

• In Event: AI, (Dis)information, and Public Accountability

Abstract

A variety of industries and firms have adopted AI ethics commitments to increase consumer trust and mitigate risks. Oversight of AI applications, however, is not uniform. The human-centered regulatory regime present in the EU differs significantly from the free-market approach proffered in the United States. While prior work has considered the role of ethics commitments in private governance, this paper provides empirical evidence on the effectiveness of AI audits in swaying public perception and the role of information in shaping consumer preferences on AI audits. We administered a survey experiment to 3,002 participants in the U.S. and Germany. The survey first assessed the role of information about AI audits in shaping consumer opinions through the random assignment of an informational video, to determine if consumers’ opinions depend on their knowledge about AI audits. Second, we use a choice-based conjoint design to explore whether factors like the nature of the audit, the auditing organization, and characteristics of the audited company matter for consumers. We pool the samples from the U.S. and Germany to determine the effect for each attribute. Then, we analyze heterogeneous effects between the US and German samples. We find consistent evidence that firms are rewarded for performing audits, that any company scandal is detrimental for respondents’ attitudes toward the firm, and that company self-audits are less valued than external audits. This paper adds to the growing literature on AI governance by providing empirical measures of the influence of AI audits on consumers’ trust of—and regulatory beliefs toward—a company.

Authors

• Alexander Wilhelm, Purdue University
  Presenting Author

• Daniel S. Schiff, Purdue University
  Non-Presenting Co-Author

• Kaylyn Jackson Schiff, Purdue University
  Non-Presenting Co-Author