Learner AI over-reliance is reframed as misclassification: a failure of real-time metacognitive evaluation with two distinct pathways¶

Claim · All claims
Evidence · 1 study (3 entries) · theoretical r2–r3 · q1
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q1 i? Misclassification is defined as selecting a subzone-inappropriate AI interaction mode for a task at a given developmental stage, and implies a competency deficit rather than a norm violation. → Tsim 2026 q1 i? Pathway one (Aid treated as Substitute) arises from effort minimization and an authority heuristic, eliminating the cognitive struggle through which skills would form. → Tsim 2026 (2) q1 i? Pathway two (Aid treated as Complement) arises from Dunning-Kruger-driven overconfidence, producing an illusion of competence where plausible AI errors pass unverified. → Tsim 2026 (3)

Evidence¶

Tsim 2026¶

Tsim, F., Gutoreva, A., Weiss, A., & Dubosh, N. (2026). AI as Teammate: Rethinking Task Distribution in Medical Training. Correspondence: [email protected]. https://arxiv.org/abs/2608.28373 doi:10.48550/arxiv.2608.28373

q1 · i? · theoretical · r3

Theoretical argument in the Clinical Reasoning Development section defining misclassification as "selecting a subzone-inappropriate AI interaction mode" and contrasting it with misuse framings that collapse inappropriate use into a single behavioral pattern without psychological mechanisms.

"misclassification, which is defined as selecting a subzone-inappropriate AI interaction mode for a given task at a given stage of development"

Tsim 2026 (2)¶

Tsim, F., Gutoreva, A., Weiss, A., & Dubosh, N. (2026). AI as Teammate: Rethinking Task Distribution in Medical Training. Correspondence: [email protected]. https://arxiv.org/abs/2608.28373. doi:10.48550/arxiv.2608.28373

q1 · i? · theoretical · r2

Theoretical argument attributing the first misclassification path to effort minimization and, secondarily, an authority heuristic whereby AI presence signals epistemic authority, producing a System 1 bypass of the deliberate System 2 thinking zone identification requires.

"First, it is due to effort minimization (Misiejuk et al., 2026): a learner defaults to automation as it is the path of the least resistance."

Tsim 2026 (3)¶

Tsim, F., Gutoreva, A., Weiss, A., & Dubosh, N. (2026). AI as Teammate: Rethinking Task Distribution in Medical Training. Correspondence: [email protected]. https://arxiv.org/abs/2608.28373. doi:10.48550/arxiv.2608.28373

q1 · i? · theoretical · r3

Theoretical argument for the second path: a learner engages as a 'proficient' delegator without the task-specific knowledge Complement requires, so plausible AI errors pass through verification that feels rigorous but is not.

"This action could be attributed to overconfidence driven by the Dunning-Kruger effect (Kruger & Dunning, 1999): a learner's insufficient task-specific knowledge produces an inflated self-assessment"

Discussion¶

Related Claims¶

• Passive engagement within correctly classified Aid tasks is a detection-resistant mis-skilling pathway requiring expert epistemic auditing — related
• Task-level account of upskilling and the Triad of Skill Failure (deskilling, never-skilling, mis-skilling) with differing empirical support — a broader claim this one bears on
• Verification bottleneck: as reliance on AI increases, accuracy in detecting AI errors decreases while confidence does not — related