The "Crutch Effect"
When generative AI is integrated as a friction-free shortcut, students experience immediate grade boosts — and suffer a severe learning deficit once the tool is withdrawn. The shortcut and the deficit are the same mechanism, just measured at different times.
Neurocognitive Evidence
A growing body of research points the same direction:
- MIT Media Lab (2025) — EEG measurements indicate a marked reduction in prefrontal cortex activation when learners rely on AI for writing tasks.
- Richmond (2025) — the "Performance Paradox": high superficial performance masks an accelerating accumulation of cognitive debt.
- Mahajan (2025) — repeated cognitive offloading triggers mental atrophy, reducing independent problem-solving stamina.
- Zhu et al. (2023) — a documented decline in learner tolerance for ambiguity and complex conceptual struggle.
- Girlich (2025) — survey data establishing frequent AI reliance as a leading predictor of reduced critical reasoning abilities.
Note: citations summarised from the Queryous content track's internal literature review.
Preserving Productive Struggle
This is why prompt engineering curricula must deliberately introduce epistemic friction — teaching students to critique, audit, and deconstruct AI outputs rather than accepting them as authoritative truths on first read.