Year: 2026
Type: article
Abstract: Artificial intelligence has become a persistent layer of work and learning rather than an occasional tool. Users must repeatedly decide when to delegate, how to prompt, whether to trust, how to verify, and how responsibility should be distributed. This integrative review defines AI fatigue as a cumulative reduction in the cognitive, affective, or motivational capacity for adaptive AI engagement when interaction, evaluation, correction, learning, and role-adjustment demands accumulate faster than recovery and support. The review connects technostress, automation vigilance, cognitive load and offloading, trust calibration, digital fatigue, and the Job Demands–Resources perspective. It introduces an Exposure–Adaptation–Depletion–Response framework, a five-domain taxonomy, and an AI Fatigue Load model that separates use volume from uncertainty, task stakes, correction burden, interruption, role disruption, and imposed adoption velocity. The analysis is updated to incorporate a 2026 validated 15-item AI Fatigue Scale; the proposed load model complements that psychophysiological outcome measure by specifying socio-technical antecedents and exposure conditions. Educational and organizational applications show why AI may simultaneously improve output and worsen human sustainability. The article identifies recurring inferential errors, presents a measurement and longitudinal research agenda, and proposes an intervention hierarchy spanning interface design, workflow, curriculum, organizational policy, and governance. AI fatigue is positioned neither as generic screen tiredness nor as a synonym for burnout, but as an adaptation problem located in the coupled human–AI–institution system.
DOI: 10.54878/bxxewh81
Author: Wael Badawy
Institution: Egyptian Russian University
Language: English
Subfield: Social Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W7223198136