Psychological pathways to responsible generative AI decision-making among teachers: governance clarity, perceived accountability, decision uncertainty, and AI literacy

Introduction Generative artificial intelligence (GenAI) can assist teachers with organizing information, generating alternatives, and drafting feedback. Its use in consequential judgments requires verifiable evidence and professional oversight. This study examines how teachers regulate optional GenAI assistance while preserving professional judgment. Methods Research on felt accountability, metacognitive monitoring, and AI literacy is conceptually integrated to develop a psychological process model of teacher decision episodes. Governance clarity is specified through rule clarity and role clarity, with responsible GenAI decision-making defined as regulated decision conduct. Results The conceptual analysis produces five testable propositions and one configurational expectation. Governance clarity is proposed to support responsible conduct through perceived accountability. Decision uncertainty is modeled as a metacognitive signal whose behavioral value depends on calibration and regulation; AI literacy is specified as a resource for verification, calibrated reliance, and appropriate escalation. Responsible conduct encompasses appropriate delegation, verification, calibrated reliance and accountable human authority, traceability, and escalation or override. These model outputs are theoretical propositions, not empirically established effects. Discussion The framework distinguishes generative assistance from conventional algorithmic advice and identifies task-dependent applications in assessment, feedback, referral, and student support. Future tests should compare unaided and assisted judgment, distinguish process quality from decision accuracy, and examine stakes, reversibility, and contextual dependence. Teachers retain authority to accept, modify, or reject GenAI output, or proceed without GenAI assistance.