Search results

Abstract
As artificial intelligence (AI) policy has begun to take shape in recent years, policy actors have worked to influence policymakers by strategically promoting issue frames that define the problems and solutions policymakers should attend to. Three such issue frames are especially prominent, surrounding AI's economic, geopolitical, and ethical dimensions. Relatedly, while technology policy is traditionally expert-dominated, new governance paradigms are encouraging increased public participation along with heightened attention to social and ethical dimensions of technology. This study aims to provide insight into whether members of the public and the issue frames they employ shape—or fail to shape—policymaker agendas, particularly for highly contested and technical policy domains. To assess this question, the study draws on a dataset of approximately five million Twitter messages from members of the public related to AI, as well as corresponding AI messages from the 115th and 116th US Congresses. After using text analysis techniques to identify the prevalence of issue frames, the study applies autoregressive integrated moving average and vector autoregression modeling to determine whether issue frames used by the public appear to influence the subsequent messaging used by federal US policymakers. Results indicate that the public does lead policymaker attention to AI generally. However, the public does not have a special role in shaping attention to ethical implications of AI, as public influence occurs only when the public discusses AI's economic dimensions. Overall, the results suggest that calls for public engagement in AI policy may be underrealized and potentially circumscribed by strategic considerations.

AbstractThe policy agenda is currently being established for artificial intelligence (AI), a domain marked by complex and sweeping implications for economic transformation tempered by concerns about social and ethical risks. This article reviews the United States national AI policy strategy through extensive qualitative and quantitative content analysis of 63 strategic AI policy documents curated by the federal government between 2016 and 2020. Drawing on a prominent theory of agenda setting, the Multiple Streams Framework, and in light of competing paradigms of technology policy, this article reviews how the U.S. government understands the key policy problems, solutions, and issue frames associated with AI. Findings indicate minimal attention to focusing events or problem indicators emphasizing social and ethical concerns, as opposed to economic and geopolitical ones. Further, broad statements noting ethical dimensions of AI often fail to translate into specific policy solutions, which may be explained by a lack of technical feasibility or value acceptability of ethics‐related policy solutions, along with institutional constraints for agencies in specific policy sectors. Finally, despite widespread calls for increased public participation, proposed solutions remain expert dominated. Overall, while the emerging U.S. AI policy agenda reflects a striking level of attention to ethics—a promising development for policy stakeholders invested in AI ethics and more socially oriented approaches to technology governance—this success is only partial and is ultimately layered into a traditional strategic approach to innovation policy.

AbstractAre narratives as influential in gaining the attention of policymakers as expert information, including for complex, technical policy domains such as artificial intelligence (AI) policy? This pre‐registered study uses a field experiment to evaluate legislator responsiveness to policy entrepreneur outreach. In partnership with a leading AI think tank, we send more than 7300 U.S. state legislative offices emails about AI policy containing an influence strategy (providing a narrative, expert information, or the organization's background), along with a prominent issue frame about AI (emphasizing technological competition or ethical implications). To assess engagement, we measure link clicks to further resources and webinar registration and attendance. Although AI policy is a highly technical domain, we find that narratives are just as effective as expert information in engaging legislators. Compared to control, expert information and narratives led to 28 and 34 percent increases in policymaker engagement, respectively. Furthermore, higher legislature professionalism and lower state‐level prior AI experience are associated with greater engagement with both narratives and expert information. Finally, we find that policymakers are equally engaged by an ethical framing of AI policy as they are with an economic one. The findings advance efforts to bridge scholarship on policy narratives, policy entrepreneurship, and agenda‐setting.

AbstractIn the context of rising delegation of administrative discretion to advanced technologies, this study aims to quantitatively assess key public values that may be at risk when governments employ automated decision systems (ADS). Drawing on the public value failure framework coupled with experimental methodology, we address the need to measure and compare the salience of three such values—fairness, transparency, and human responsiveness. Based on a preregistered design, we administer a survey experiment to 1460 American adults inspired by prominent ADS applications in child welfare and criminal justice. The results provide clear causal evidence that certain public value failures associated with artificial intelligence have significant negative impacts on citizens' evaluations of government. We find substantial negative citizen reactions when fairness and transparency are not realized in the implementation of ADS. These results transcend both policy context and political ideology and persist even when respondents are not themselves personally impacted.

This study addresses the phenomenon of misinformation about misinformation, or politicians "crying wolf" over fake news. Strategic and false claims that stories are fake news or deepfakes may benefit politicians by helping them maintain support after a scandal. We posit that this benefit, known as the "liar's dividend," may be achieved through two politician strategies: by invoking informational uncertainty or by encouraging oppositional rallying of core supporters. We administer five survey experiments to over 15,000 American adults detailing hypothetical politician responses to stories describing real politician scandals. We find that claims of misinformation representing both strategies raise politician support across partisan subgroups. These strategies are effective against text-based reports of scandals, but are largely ineffective against video evidence and do not reduce general trust in media. Finally, these false claims produce greater dividends for politicians than alternative responses to scandal, such as remaining silent or apologizing.

AbstractLaw enforcement agencies are increasingly adopting artificial intelligence (AI)‐powered tools. While prior work emphasizes the technological features driving public opinion, we investigate how public trust and support for AI in government vary with the institutional context. We administer a pre‐registered survey experiment to 4200 respondents about AI use cases in policing to measure responsiveness to three key institutional factors: bureaucratic proximity (i.e., local sheriff versus national Federal Bureau of Investigation), algorithmic targets (i.e., public targets via predictive policing versus detecting officer misconduct through automated case review), and agency capacity (i.e., necessary resources and expertise). We find that the public clearly prefers local over national law enforcement use of AI, while reactions to different algorithmic targets are more limited and politicized. However, we find no responsiveness to agency capacity or lack thereof. The findings suggest the need for greater scholarly, practitioner, and public attention to organizational, not only technical, prerequisites for successful government implementation of AI.

Abstract
Designing effective and inclusive governance and public communication strategies for artificial intelligence (AI) requires understanding how stakeholders reason about its use and governance. We examine underlying factors and mechanisms that drive attitudes toward the use and governance of AI across six policy-relevant applications using structural equation modeling and surveys of both US adults (N = 3,524) and technology workers enrolled in an online computer science master's degree program (N = 425). We find that the cultural values of individualism, egalitarianism, general risk aversion, and techno-skepticism are important drivers of AI attitudes. Perceived benefit drives attitudes toward AI use but not its governance. Experts hold more nuanced views than the public and are more supportive of AI use but not its regulation. Drawing on these findings, we discuss challenges and opportunities for participatory AI governance, and we recommend that trustworthy AI governance be emphasized as strongly as trustworthy AI.