Representing technological "minds": How anthropomorphic inferences influence legal judgments and policy opinions

The increasing presence of autonomous and semi-autonomous technologies is raising challenging issues of law and policy. How society resolves these issues will likely depend in part on how its leaders and members conceptualize these technologies. While research indicates that people often draw anthropomorphic inferences about these technologies, we know little about how these inferences influence technology-related legal judgments and policy opinions. We report three studies in which participants read narratives about accidents involving autonomous machines, presented with or without anthropomorphic framing, then answered questions concerning their attributions to the machines, their experience of cognitive conflict, and their legal judgments and policy opinions. We found that participants were far more willing to attribute human-like agency than human-like experience to the machines, and these different types of attributions had distinct and sometimes opposing influences on participants' judgments. For instance, attributions of agency to machines were associated with decreased beliefs that the machines' owners or manufacturers should be liable for harms the machines cause, while attributions of experience were associated with increased beliefs. More broadly, we found that increases in participants' internal cognitive conflict predicted increases in their experience attributions, which, in turn, predicted (i) more autonomous-technology-friendly policy opinions and (ii) greater willingness to impose legal liability when those technologies cause harm. Finally, framing had minimal influence on participants' responses. Our findings illuminate ways in which anthropomorphic attributions can influence judgments in an important and socially relevant context.