Trust and trustworthy artificial intelligence: A research agenda for AI in the environmental sciences.

Demands to manage the risks of artificial intelligence (AI) are growing. These demands and the government standards arising from them both call for trustworthy AI. In response, we adopt a convergent approach to review, evaluate, and synthesize research on the trust and trustworthiness of AI in the environmental sciences and propose a research agenda. Evidential and conceptual histories of research on trust and trustworthiness reveal persisting ambiguities and measurement shortcomings related to inconsistent attention to the contextual and social dependencies and dynamics of trust. Potentially underappreciated in the development of trustworthy AI for environmental sciences is the importance of engaging AI users and other stakeholders, which human-AI teaming perspectives on AI development similarly underscore. Co-development strategies may also help reconcile efforts to develop performance-based trustworthiness standards with dynamic and contextual notions of trust. We illustrate the importance of these themes with applied examples and show how insights from research on trust and the communication of risk and uncertainty can help advance the understanding of trust and trustworthiness of AI in the environmental sciences.

Enabling conditions for an equitable and sustainable blue economy.

The future of the global ocean economy is currently envisioned as advancing towards a 'blue economy'-socially equitable, environmentally sustainable and economically viable ocean industries1,2. However, tensions exist within sustainable development approaches, arising from differing perspectives framed around natural capital or social equity. Here we show that there are stark differences in outlook on the capacity for establishing a blue economy, and on its potential outcomes, when social conditions and governance capacity-not just resource availability-are considered, and we highlight limits to establishing multiple overlapping industries. This is reflected by an analysis using a fuzzy logic model to integrate indicators from multiple disciplines and to evaluate their current capacity to contribute to establishing equitable, sustainable and viable ocean sectors consistent with a blue economy approach. We find that the key differences in the capacity of regions to achieve a blue economy are not due to available natural resources, but include factors such as national stability, corruption and infrastructure, which can be improved through targeted investments and cross-scale cooperation. Knowledge gaps can be addressed by integrating historical natural and social science information on the drivers and outcomes of resource use and management, thus identifying equitable pathways to establishing or transforming ocean sectors1,3,4. Our results suggest that policymakers must engage researchers and stakeholders to promote evidence-based, collaborative planning that ensures that sectors are chosen carefully, that local benefits are prioritized, and that the blue economy delivers on its social, environmental and economic goals.

Indiscriminate, Irrelevant, and Sometimes Wrong: Causal Misconceptions about Climate Change.

Prior research demonstrates widespread persistence of beliefs about climate change causes and risks that are arguably misconceptions. They include believing pollution causes climate change, believing ozone depletion causes climate change, the combination of these two "green beliefs," referred to as environmental problems, and believing natural climate variation significantly contributes to current climate trends. Each of these causal beliefs has the potential to weaken or divert support away from effective climate change risk mitigation policies. To assess this potential, we explore the nature and prevalence of these beliefs in the United States with a national sample of interviews (N = 77) and two national surveys (N = 1,013, N = 1,820), and apply regression and mediation analyses to explore whether they explain any of the variation in individuals' concern or support for policy to mitigate climate change. Adherence to these beliefs-which reflect a variety of misconceptions illustrated in the interviews-differs by political ideology but is common, with over a third of interviewees mentioning one or more. Controlling for general knowledge, political ideology, and other factors, misconceptions about environmental problems are still associated directly with support for climate change policies. On average adherence to the belief that environmental problems cause climate change is associated with a 25% higher probability of policy support. In contrast, believing natural climate variability is a major recent cause of climate change is associated with a 7% lower probability of supporting climate policy, even after controlling for political ideology and other knowledge about climate change.

Efficacy Foundations for Risk Communication: How People Think About Reducing the Risks of Climate Change.

Believing action to reduce the risks of climate change is both possible (self-efficacy) and effective (response efficacy) is essential to motivate and sustain risk mitigation efforts, according to current risk communication theory. Although the public recognizes the dangers of climate change, and is deluged with lists of possible mitigative actions, little is known about public efficacy beliefs in the context of climate change. Prior efficacy studies rely on conflicting constructs and measures of efficacy, and links between efficacy and risk management actions are muddled. As a result, much remains to learn about how laypersons think about the ease and effectiveness of potential mitigative actions. To bring clarity and inform risk communication and management efforts, we investigate how people think about efficacy in the context of climate change risk management by analyzing unprompted and prompted beliefs from two national surveys (N = 405, N = 1,820). In general, respondents distinguish little between effective and ineffective climate strategies. While many respondents appreciate that reducing fossil fuel use is an effective risk mitigation strategy, overall assessments reflect persistent misconceptions about climate change causes, and uncertainties about the effectiveness of risk mitigation strategies. Our findings suggest targeting climate change risk communication and management strategies to (1) address gaps in people's existing mental models of climate action, (2) leverage existing public understanding of both potentially effective mitigation strategies and the collective action dilemma at the heart of climate change action, and (3) take into account ideologically driven reactions to behavior change and government action framed as climate action.

Efficacy, Action, and Support for Reducing Climate Change Risks.

A growing body of research demonstrates that believing action to reduce the risks of climate change is both possible (self-efficacy) and effective (response efficacy) is essential to motivate and sustain risk mitigation efforts. Despite this potentially critical role of efficacy beliefs, measures and their use vary wildly in climate change risk perception and communication research, making it hard to compare and learn from efficacy studies. To address this problem and advance our understanding of efficacy beliefs, this article makes three contributions. First, we present a theoretically motivated approach to measuring climate change mitigation efficacy, in light of diverse proposed, perceived, and previously researched strategies. Second, we test this in two national survey samples (Amazon's Mechanical Turk N = 405, GfK Knowledge Panel N = 1,820), demonstrating largely coherent beliefs by level of action and discrimination between types of efficacy. Four additive efficacy scales emerge: personal self-efficacy, personal response efficacy, government and collective self-efficacy, and government and collective response efficacy. Third, we employ the resulting efficacy scales in mediation models to test how well efficacy beliefs predict climate change policy support, controlling for specific knowledge, risk perceptions, and ideology, and allowing for mediation by concern. Concern fully mediates the relatively strong effects of perceived risk on policy support, but only partly mediates efficacy beliefs. Stronger government and collective response efficacy beliefs and personal self-efficacy beliefs are both directly and indirectly associated with greater support for reducing the risks of climate change, even after controlling for ideology and causal beliefs about climate change.