Whose metaphor is it anyway? Analysing AI metaphors from positionality and values of speaker and recipient

Abstract

This article builds on our earlier work on metaphors of generative AI as a route to developing critical AI literacy (Gupta, et al., 2024). Using a duo-autoethnographic approach, we analyse metaphors we have encountered in workshops, classrooms, and informal conversations. We reflect on how our positionalities shape how we receive and interpret these metaphors. Where possible, we also consider the context and likely values of the speakers who used these metaphors, while recognising that our study does not include formal interviews and therefore cannot fully recover speakers’ values. Rather than theorising metaphors in abstract terms, we examine the ethical and affective dimensions of specific examples, including “AI is stupid” and “AI as cane.” Our analysis shows how such metaphors can reinforce or resist ableism, epistemic injustice, and uncritical adoption of AI. We conclude by suggesting classroom practices that invite students to surface and critically examine their own metaphors for AI.