Abstract The rhetoric of “human‐AI collaboration” frames contemporary discussions of interacting with artificial intelligence systems. Drawing from 15 years of research in collaborative information seeking, we argue that this metaphor fundamentally misrepresents these interactions and obscures critical issues of agency, accountability, and labor. Real collaboration—whether mediated by algorithms or not—requires mutual understanding, shared goals, reciprocal adaptation, and common ground building. Current AI systems meet none of these criteria. Here we examine what genuine collaboration entails, why the AI industry's appropriation of this language matters, and what we lose when we mistake sophisticated autocomplete for partnership.
Shah et al. (Thu,) studied this question.