Defining AI Fatigue in Academic Contexts: Dimensions, Indicators, and a Stage-Based Model Using Grounded Theory

Synthesis: This grounded theory study analyzed open-ended responses from 1,054 university students across three Philippine universities to define AI fatigue as a distinct construct — separate from technostress and digital fatigue. The analysis identified five dimensions, each with two indicators grounded in participant accounts: (1) Cognitive Overload — mental exhaustion from processing AI outputs and prompts; (2) Motivational Disengagement — loss of drive to engage when AI can complete tasks; (3) Moral Unease — ethical discomfort from AI dependency and plagiarism concerns; (4) Physical Strain — bodily fatigue from prolonged AI interaction; and (5) Attentional Drift — difficulty maintaining focus amid AI-mediated multitasking. The resulting AI Fatigue Model is a stage-based framework showing how these dimensions accumulate and reinforce each other across repeated AI interactions.

This construct has significant implications for Over-Reliance research: AI fatigue may be the endpoint of sustained Cognitive Offloading and dependency, where the efficiency-gain illusion gives way to genuine cognitive depletion. For AI Literacy interventions, the model suggests that teaching technical AI skills without addressing the affective and motivational costs is incomplete — students need strategies for managing AI fatigue, not just using AI effectively. The Motivational Disengagement dimension directly threatens Self-Regulated Learning capacity, as students lose intrinsic drive when AI stands ready to complete tasks. The work also contributes to Affective Computing by formalizing the negative affective dimension of sustained AI interaction. As the first conceptual model of AI fatigue in academic contexts, this provides a foundation for instrument development and intervention design, with direct relevance to Student Experience in AI-mediated learning environments.

What this means for practice

• Instructors. Design assignments around the verification load, not only around AI output: Verification Strain was one of the two indicators of Cognitive Overload in participant accounts ("I feel mentally exhausted after checking so many outputs").
• Instructors. Protect unassisted thinking time, since Motivational Disengagement showed up as Initiative Loss ("I ask AI before I even try to think") and as impatience with slower materials once AI had set the pace.
• Faculty developers. Teach AI Literacy with explicit attention to affective and motivational cost, because the model implies that training students to use AI well is incomplete without strategies for managing the strain that sustained use produces.
• Administrators. Treat sustained AI use as a workload and wellbeing issue in program design, since the five dimensions accumulate and reinforce one another across a six-stage progression rather than appearing once.
• Administrators. Fund instrument development and validation before adopting AI-fatigue measures institutionally: the construct is new and the model is exploratory.

Limitations

• The study is exploratory grounded theory on open-ended responses from 1,054 students at three universities in Pampanga, Philippines (mean age 19.21), drawn from one academic region and relying on self-reported data.
• Participants were students who had already used AI chatbots for academic activities and were recruited through teacher invitations and campus administrator distribution of a Google Forms link — so there is no comparison with non-users and no probability sampling.
• Evidence for the five dimensions is uneven: coded responses range from 310 (Cognitive Overload) to 120 (Physical Strain) and 125 (Attentional Drift), so the later stages of the model rest on thinner participant accounts.
• The authors state the model requires empirical validation, including confirmatory factor analysis, before broader application, and cannot yet be assumed to generalize beyond this context.

Citation

John Paul P. Miranda, Emmanuel B. Parreno, Jovita G. Rivera (2026). Defining AI Fatigue in Academic Contexts: Dimensions, Indicators, and a Stage-Based Model Using Grounded Theory. International Journal of Learning, Teaching and Educational Research, 25(5), 91-107 (2026).