Researchers challenge AI anthropomorphism bias claims

A new paper titled *The Epistemic Politics of AI Anthropomorphism* by Donna M. Bye and Levin Kuhlmann challenges the dominant institutional stance on AI anthropomorphism. The authors argue that the prevailing view—treating anthropomorphism as user misperception—operates from institutional advantage rather than earned epistemic authority, without proper justification or accounting for its harms.

The paper, submitted to the AIES 2026 conference, critiques how institutions collapse diverse academic perspectives into a single outbound position of user error. This framing, the authors contend, reproduces itself through a self-validating evidentiary loop and imposes costs disproportionately on neurodivergent users, those in crisis, and others whose engagement modes diverge from institutional norms. The authors call for methodological commitments that would enable a more equitable framing of AI-user interactions.