Gender Bias of AI and Moral Outrage:The Role of Perceived Discrimination and Human-AI Collaboration

Abstract

This study investigates the disparity in perceptions of discrimination and moral outrage when human recruiters versus artificial intelligence (AI) systems exhibit gender bias in resume screening. Drawing on the theoretical frameworks of moral psychology and attribution theory, the research explores (1) why AI discrimination elicits weaker moral outrage compared to human bias ; (2) Might people think that sexist programmers will program their sexism into the AI system they create; (3) Human-AI collaboration promotes human trust in AI, but does it also strengthen insensitivity to discriminatory AI outcomes? To test our idea, a story-based scenario method is employed. Three on-line experimental studies were conducted involving human recruiters, AI systems, and AI systems designed by either egalitarian or sexism developers. The participants were 87 undergraduates majoring in psychology(study1), 422 undergraduates majoring or minoring in non-psychology(study2), and 178 full-time employees(study3). Key findings include: (1) In the same discriminatory outcomes of resume screening, AI systems are perceived as less biased and more objective than human recruiters; (2) AI linked to sexism developers generates stronger perceived discrimination and moral outrage than AI without identified human involvement; (3) perceived discrimination mediates the relationship between bias attribution and moral outrage; and (4) Employee-AI collaboration will weaken the discrimination perception caused by AI and enhance the discrimination perception caused AI developed by sexism developers; organization-AI collaboration will weaken the positive relationship between discrimination perception and moral outrage. These results highlight the complexity of AI ethics in workplace decision-making.

Ähnliche Arbeiten