Structural explanations reduce moral outrage toward AI but not human ethnic discrimination in credit lending

When AI-based decision-making systems discriminate against particular groups of individuals, this discrimination tends to elicit less moral outrage than when humans discriminate, a phenomenon referred to as the algorithmic outrage deficit. The present research replicates and extends this effect to ethnicity-based discrimination (based on migration background) in credit lending. Participants in our study morally evaluated scenarios of discriminatory credit lending decisions made by either a bank employee or an AI. Results support the algorithmic outrage deficit: Participants expressed less moral outrage toward AI than toward human discrimination, mediated by an attenuated attribution of biased motives to the AI. Extending prior research, we also tested how providing structural explanations for why the discrimination occurred (biased training data/past experiences reflecting historical injustice) affected responses. Results show that structural explanations (and self-reported knowledge about AI systems) further reduced algorithmic outrage. These findings suggest that the algorithmic outrage deficit is a broad phenomenon and that transparency about the structural origins of algorithmic bias may paradoxically further reduce moral outrage.