Trusting the Moral Judgments of a Robot: Perceived Moral Competence and Humanlikeness of a GPT-3 Enabled AI

Abstract

Advancements in computing power and foundational modeling have enabled artificial intelligence (AI) to respond to moral queries with surprising accuracy. This raises the question of whether we trust AI to influence human moral decision-making, so far, a uniquely human activity. We explored how a machine agent trained to respond to moral queries (Delphi, Jiang et al., 2021) is perceived by human questioners. Participants were tasked with querying the agent with the goal of figuring out whether the agent, presented as a humanlike robot or a web client, was morally competent and could be trusted. Participants rated the moral competence and perceived morality of both agents as high yet found it lacking because it could not provide justifications for its moral judgments. While both agents were also rated highly on trustworthiness, participants had little intention to rely on such an agent in the future. This work presents an important first evaluation of a morally competent algorithm integrated with a human-like platform that could advance the development of moral robot advisors.