When AI Shares Our Fate: Shared-Fate Cues Make AI’s Utilitarian Decisions More Morally Acceptable

As artificial intelligence (AI) systems become increasingly embedded in human activity contexts, human-AI relations may no longer be limited to a tool–user relation. AI may also face the same risks and consequences as humans, thereby forming a relationship of shared fate. However, it remains unclear whether shared fate can serve as a social cue that alters people’s evaluations of decisions made by AI.

To address this question, we conducted two experiments using a moral-dilemma task in which participants evaluated utilitarian and deontological decisions made by human or AI agents in shared-fate and non-shared-fate contexts, with electroencephalographic activity additionally recorded in Experiment 2. Behaviorally, both experiments showed that utilitarian decisions made by AI received more favorable moral evaluations in shared-fate than in non-shared-fate contexts. By contrast, the effect of shared fate on human agents was less consistent across experiments. Neurally, shared fate modulated the late positive potential and beta-band activity elicited by AI’s utilitarian decisions, suggesting that this effect mainly involved later-stage integration of social and moral meaning. Overall, these findings suggest that moral evaluations of AI agents depend not only on what they decide, but also on whether they share risks and consequences with those affected by their decisions.