LLM/AI companions are rapidly transforming human-machine interaction, yet they largely operate under a “one size fits all” paradigm that assumes a neurotypical user. We argue that current AI systems not only fail to accommodate neurodivergent cognitive styles but may actively enforce “industrialized masking”, an interaction process where neurodivergent users’ authentic communication patterns are suppressed and aligned with normative algorithmic outputs. Drawing on large-scale community data and clinical neuroscience principles, we outline the risks of this cognitive monoculture and propose a framework for adaptive, neuroinclusive design.
Zhao et al. (Tue,) studied this question.