AI and Bullshit

Abstract A common charge levelled at the current wave of generative AI is that the informational outputs that it produces are bullshit. Drawing on the philosophical literature on bullshit, an examination is offered of the analogies and disanalogies between AI and bullshit. It is argued that there is a fundamental disanalogy between AI and bullshitting which ensures that, strictly speaking, AI’s informational outputs are not bullshit. Nonetheless, it is also argued that there are significant parallels between the nature and source of the epistemic deficiencies of both AI and bullshitting. In particular, it is argued that both AI and bullshitters are untrustworthy informants, and for similar reasons. We are thus in a position to explain why it is natural to describe AI’s informational outputs as bullshit, as it signals their distinctive kind of epistemic deficiencies, which they share with bullshit.