LLMs as Narrators of Lived Experience: A Study of Generative Hermeneutical Epistemic Injustice in ADHD Sense-Making

Abstract

Large language models (LLMs), as non-human actors, do not have access to first-hand experiences. Yet they are increasingly consulted in processes of making sense of them, from mental health chatbots to reflexive journals. As a result, complex lived realities risk being distorted or flattened through second-order representations. In this paper, we expand on conceptual work to empirically examine narrative mechanisms through which LLMs as narrators give rise to hermeneutical epistemic injustice. We foreground narration as a key interpretive process through which experience is made intelligible. Focusing on ADHD as a motivating case, we compare first-person narratives from online ADHD communities with narratives generated by LLMs. Using an interpretive, reflexive analysis, we examine how these forms of narration construct lived experience. We find that generated narratives organise ADHD experience through recurring modes that individualise, standardise, and stabilise interpretation, reducing messiness, complexity and ambiguity present in community narration. More situated and unresolved accounts remained possible when foregrounded through prompting. We argue that the epistemic concern lies not only in what models can represent, but in which interpretations they make readily available and which require prior conceptual and prompting work.