A novel dynamical systems model predicts a critical AI capability threshold, beyond which human skills abruptly collapse due to delegation, a phenomenon termed the "enrichment paradox." The model suggests that periodic AI failures and mandatory practice can paradoxically preserve human capabilities and highlights the irreversible nature of profound skill loss.