Can We Place Epistemic Trust in AI?

As AI systems become an important part of our life, theoretical investigations into AI are emerging. It is pressing to ask in what sense we trust AI. Many scholars have expressed concerns regarding moral trust in AI, while epistemic trust in AI seems to offer a novel option. Reflecting on McCraw’s and Alvarado’s proposal, we conclude that epistemic trust in AI cannot be distinguished from epistemic reliance on AI. An indirect line considers whether we can epistemically blame AI. Having examined three views of epistemic blame, we find that none of them can make a case for epistemic blame on AI, and therefore we are led to think that we can’t have epistemic trust in AI in the first place. As a result, current theories cannot establish that we can place epistemic trust in AI.