To err is human: Bias salience can help overcome resistance to medical AI

Abstract

Prior research has shown that many individuals exhibit an aversion to algorithms and are resistant to the use of artificial intelligence (AI) in healthcare. In the present research, we show that an intervention that increases the salience of bias in decision making—either in general or specifically with respect to gender or age—makes individuals relatively more receptive to medical AI. This increased receptiveness to AI occurs because bias is perceived to be a fundamentally human shortcoming. As such, when the prospect of bias is made salient, perceptions of AI integrity—defined as the perceived fairness and trustworthiness of an AI agent relative to a human counterpart—are enhanced.