Generative AI, especially chatbots, are being incorporated into many epistemic domains such as schooling and research. Indeed, Ramón Alvarado (2023) argues artificial intelligence should be understood as an epistemic technology. In many of these domains, we’re being asked to extend a kind of epistemic trust to generative AI chatbots.

How, then, should we think about trust from a feminist perspective when it comes to chatbots? Drawing on feminist epistemologies, we first turn to the literature for a richer account of competence which takes seriously the ways in which who one is can affect the knowledge that one has, and thus offer to others. Knowers are fundamentally situated and one’s situatedness affects one’s contextually-specific epistemic trustworthiness. We need to judge if a person is well-placed to draw accurate conclusions in this context; we need to know what concepts they work with, the biases they have, their interpretation of the background information, and their projectability judgements. We argue there are systematic challenges in making these kinds of trust judgements because of the barriers to assessing the epistemic positioning of a chatbot. Our analysis has practical implications for how and where we use chatbots.