Surveying Public Perceptions of Artificial Intelligence in Health Care in the United States: Systematic Review.

BACKGROUND: This paper reviews nationally representative public opinion surveys on artificial intelligence (AI) in the United States, with a focus on areas related to health care. The potential health applications of AI continue to gain attention owing to their promise as well as challenges. For AI to fulfill its potential, it must not only be adopted by physicians and health providers but also by patients and other members of the public. OBJECTIVE: This study reviews the existing survey research on the United States' public attitudes toward AI in health care and reveals the challenges and opportunities for more effective and inclusive engagement on the use of AI in health settings. METHODS: We conducted a systematic review of public opinion surveys, reports, and peer-reviewed journal articles published on Web of Science, PubMed, and Roper iPoll between January 2010 and January 2022. We include studies that are nationally representative US public opinion surveys and include at least one or more questions about attitudes toward AI in health care contexts. Two members of the research team independently screened the included studies. The reviewers screened study titles, abstracts, and methods for Web of Science and PubMed search results. For the Roper iPoll search results, individual survey items were assessed for relevance to the AI health focus, and survey details were screened to determine a nationally representative US sample. We reported the descriptive statistics available for the relevant survey questions. In addition, we performed secondary analyses on 4 data sets to further explore the findings on attitudes across different demographic groups. RESULTS: This review includes 11 nationally representative surveys. The search identified 175 records, 39 of which were assessed for inclusion. Surveys include questions related to familiarity and experience with AI; applications, benefits, and risks of AI in health care settings; the use of AI in disease diagnosis, treatment, and robotic caregiving; and related issues of data privacy and surveillance. Although most Americans have heard of AI, they are less aware of its specific health applications. Americans anticipate that medicine is likely to benefit from advances in AI; however, the anticipated benefits vary depending on the type of application. Specific application goals, such as disease prediction, diagnosis, and treatment, matter for the attitudes toward AI in health care among Americans. Most Americans reported wanting control over their personal health data. The willingness to share personal health information largely depends on the institutional actor collecting the data and the intended use. CONCLUSIONS: Americans in general report seeing health care as an area in which AI applications could be particularly beneficial. However, they have substantial levels of concern regarding specific applications, especially those in which AI is involved in decision-making and regarding the privacy of health information.

How institutional factors at US land-grant universities impact scientists' public scholarship.

The call for public scholarship to emphasize the broader impacts of science has raised questions about how universities can support this work among their scientists. This study quantitatively assesses how institutional factors shape scientists' participation in public scholarship, a subset of public engagement focusing on scientists' involvement in public debate and democratic decision-making related to science policy. Based on a 2018 survey of scientists from 46 US land-grant universities (N = 6,242), hierarchical linear modeling results show that institutional factors, including tenure guidelines and the extent of government funding, play a minor role in influencing scientists' public scholarship participation. More importantly, scientists' perceptions of the university climate on support for engagement, including support from high-level administrators and for graduate students, are significant predictors of participation in public scholarship. Ultimately, these findings support the recommendation that universities should coordinate individual motivations with institutional missions to support a broader culture of public engagement.

Collateral damage from debunking mRNA vaccine misinformation.

Amid the COVID-19 pandemic, the scientific community has been understandably eager to combat misinformation about issues such as vaccine safety. In highly polarized information environments, however, even well-intentioned messages have the potential to produce adverse effects. In this study, we connect different disciplinary strands of social science to derive and experimentally test the novel hypothesis that although particular efforts to debunk misinformation about mRNA vaccines will reduce relevant misperceptions about that technology, these correctives will harm attitudes toward other types of vaccines. We refer to this as the "collateral damage hypothesis." Our study specifically examines a corrective message stating that "mRNA vaccines do not contain live virus," and our results offer some support for our hypothesis, with the corrective triggering increased societal risk perceptions of live vaccines. We also find that the effect is, predictably, most evident among those whose vaccine acceptance is low. Building on the theoretical grounding we outline, we test a "damage control" adjustment to the corrective message and present evidence supporting that it mitigates the collateral damage.

A triangulated approach for understanding scientists' perceptions of public engagement with science.

Scientists are expected to engage with the public, especially when society faces challenges like the COVID-19 pandemic or climate change, but what public engagement means to scientists is not clear. We use a triangulated, mixed-methods approach combining survey and focus group data to gain insight into how pre-tenure and tenured scientists personally conceptualize public engagement. Our findings indicate that scientists' understanding of public engagement is similarly complex and diverse as the scholarly literature. While definitions and examples of one-way forms of engagement are the most salient for scientists, regardless of tenure status, scientists also believe public engagement with science includes two-way forms of engagement, such as citizen and community involvement in research. These findings suggest that clear definitions of public engagement are not necessarily required for its application but may be useful to guide scientists in their engagement efforts, so they align with what is expected of them.

Public engagement: Faculty lived experiences and perspectives underscore barriers and a changing culture in academia.

The idea of faculty engaging in meaningful dialogue with different publics instead of simply communicating their research to interested audiences has gradually morphed from a novel concept to a mainstay within most parts of the academy. Given the wide variety of public engagement modalities, it may be unsurprising that we still lack a comprehensive and granular understanding of factors that influence faculty willingness to engage with public audiences. Those nuances are not always captured by quantitative surveys that rely on pre-determined categories to assess scholars' willingness to engage. While closed-ended categories are useful to examine which factors influence the willingness to engage more than others, it is unlikely that pre-determined categories comprehensively represent the range of factors that undermine or encourage engagement, including perceptual influences, institutional barriers, and scholars' lived experiences. To gain insight into these individual perspectives and lived experiences, we conducted focus group discussions with faculty members at a large midwestern land-grant university in the United States. Our findings provide context to previous studies of public engagement and suggest four themes for future research. These themes affirm the persistence of institutional barriers to engaging with the public, particularly the expectations in the promotion process for tenure-track faculty. However, we also find a perception that junior faculty and graduate students are challenging the status quo by introducing a new wave of attention to public engagement. This finding suggests a "trickle-up" effect through junior faculty and graduate students expecting institutional support for public engagement. Our findings highlight the need to consider how both top-down factors such as institutional expectations and bottom-up factors such as graduate student interest shape faculty members' decisions to participate in public engagement activities.

What Do We (Not) Know About Global Views of Human Gene Editing? Insights and Blind Spots in the CRISPR Era.

As research on human applications of CRISPR advances, researchers, advisory bodies, and other stakeholder organizations continue calling for global public discourses and engagement to shape the development of human gene editing (HGE). Research that captures public views and tests ways for engaging across viewpoints is vital for facilitating these discourses. Unfortunately, such research lags behind advances in HGE research and applications. Here, we provide the first review of nationally representative public-opinion surveys focused on HGE to discuss limitations and remaining gaps, illustrating how these gaps hinder interpretation of existing studies. Rigorous research with proper methods for capturing representative public opinion of HGE is limited, especially in countries outside of the United States and on a global scale. The result is severely restricted understanding of even the surface level of public views concerning HGE. We identify broad areas where we need more and better research capturing public views, and describe how future surveys can help collect insights necessary for discourse and decision making on HGE.