Professional Judgment under AI-Mediated Language Interaction

AI systems are increasingly integrated into professional environments as language-based tools that summarise information, propose interpretations, and structure relevance. While these systems are formally positioned as advisory, their interaction with professional judgment introduces a distinct governance risk: the shaping of judgment conditions without corresponding shifts in accountability, auditability, or assurance. This paper identifies how AI-mediated language interaction alters professional judgment prior to decision-making, creating consequence-bearing outcomes without explicit delegation of authority or control. The analysis situates this phenomenon within established research on professional judgment, human–AI interaction, sociotechnical risk, and AI evaluation limits, demonstrating that existing professional and regulatory frameworks remain structurally blind to this class of risk.