Judgment, Delegation and Trust in AI-Mediated Scholarly Evaluation

Artificial intelligence has become embedded in the routine practices of scholarly publishing. Authors commonly rely on AI tools to refine language; journals increasingly use automated systems to flag potential integrity concerns; and reviewers are beginning to draw on large language models to support evaluation. These developments are often treated as separate and incremental adjustments. Taken together, however, they point to a more consequential change: evaluative judgment in scholarly publishing is progressively shaped by systems whose role in judgment formation is rarely explicit. This perspective considers how the combined use of AI-assisted writing, AI-based detection, and AI-supported peer review complicates procedural legitimacy, particularly when editorial authority comes to rest on probabilistic indicators rather than articulated and contestable reasons. Scholarly publishing therefore provides a useful setting for examining how AI reconfigures established relations of trust, responsibility, and governance in high-stakes evaluative processes.