Reconfiguring Empathy: Affective Outcomes of AI-Mediated Decision-Making
Eva‐Madeleine Schmidt · 2026
When an AI filter neutralizes a distressed face or a risk model marks a welfare applicant as ineligible, it does not simply modify the data but reshapes the emotional ground on which decisions are made. This research examines how AI systems influence the affective mechanisms of humans that guide decision-making. As AI becomes embedded in institutional and everyday contexts, it increasingly alters empathy, responsibility, and affective engagement. Using controlled studies in administrative decision-making and donation appeals, the dissertation shows that even brief interactions with AI advice or AI-altered faces can shift both discretionary choices and the affective conditions in which these are made, sometimes in subtle and counterintuitive ways. By positioning affect as a critical layer of human–AI interaction, this work offers guidance for designing AI systems that balance affective engagement and support humane, accountable decision-making, contributing insights that HCI is uniquely positioned to integrate across adjacent disciplines.