Judgment Formation in AI-Mediated Inquiry: A Preliminary Generative Relational Account

Abstract

This paper develops a preliminary Generative Relational (GR) account of judgment formation in AI-mediated inquiry. Its analytical object is not judgment in general and not a contest between human and artificial judgment. The paper examines how historically developed capacities for framing, discrimination, confidence calibration, selective reliance, revision, responsibility, and learning are reorganized when artificial intelligence can participate in generating options, reasons, evaluations, recommendations, and polished answers. The starting distinction is between improved task performance and the development of judgment. AI assistance can improve performance while leaving the user poorly calibrated about the quality or provenance of the resulting decision, and algorithmic advice can influence both decisions and confidence in ways conditioned by metacognitive sensitivity. The paper therefore treats judgment formation as a history-bearing, relational process whose relevant unit may be a coupled Human–AI configuration rather than an isolated individual. Subsequent sections examine the AI-mediated judgment environment, calibration and selective reliance, problem salience, aesthetic, normative, practical, and trajectory-sensitive judgment, reflexive effects of present judgment on future judgment conditions, power over judgment criteria, delegation and responsibility, practices for judgment under AI mediation, cultivation and productive cognitive friction, and the assessment of judgment development. The framework does not reserve mature judgment to humans, prescribe a universal decision procedure, or infer judgment quality from answer accuracy alone. Its purpose is to provide a revisable conceptual architecture for studying how judgment can remain calibrated, responsibility-bearing, and developmentally generative as artificial assistance becomes more capable.