Epistemic dysmorphia: a working definition

Epistemic dysmorphia is a term introduced by Anthony R Brand (2026) for a sustained, self-reinforcing distortion of students' perception of their own cognitive development, produced by fluent output, sycophantic feedback and assessment that rewards products without verifying process. It is most pronounced in students' written assignments where generative AI — most commonly a large language model (LLM) — is used in some aspect of the writing process. Its degree reflects, in part, which aspects of that process have been offloaded to the LLM, and how far each has been offloaded. It differs from neighbouring concepts such as the Dunning–Kruger effect, which arises from a lack of skill and is exposed when performance is tested. In epistemic dysmorphia, the three conditions — fluent output, sycophantic feedback, and assessment that rewards products without verifying process — compound one another: the assessment tests the product, not the capacity, so the result silently accepts and reinforces the distortion, rather than confronting and correcting it. The note also defines its educator-side counterpart, pedagogical dysmorphia: an educator's mistaken belief that an assignment still assesses what it was designed to assess. A working definition, published to date the terms ahead of a fuller account in preparation.