Transparency, neutrality, voice, and respect: How procedural fairness considerations affect AI acceptability in algorithmic societies

Public attitudes toward artificial intelligence (AI) are often explained in terms of users’ perceived costs and benefits. More recent research emphasizes societal concerns and procedural fairness, yet these different dimensions have rarely been integrated into a unified explanatory framework. This article offers the first joint domain-general test of outcome- and process-oriented approaches to explain public acceptability of AI. We argue that beliefs about procedural fairness—specifically about transparency, neutrality, voice, and respect—become particularly consequential as AI systems assume more agent-like roles, exercising limited but relevant autonomy and mediating decision-making and social ordering. We test this framework using data from a nationally representative household survey of Portuguese residents. Structural equation modeling shows that evaluative beliefs about procedural fairness are the strongest antecedents of AI acceptability, outweighing egocentric and sociotropic outcome considerations. Familiarity with AI contributes to more positive evaluative beliefs overall but is unevenly distributed across sociodemographic groups, making it a source of social divides in AI legitimacy. Overall, the findings urge incorporation of procedural integrity as a central foundation for the public legitimacy of artificial intelligence.