Rédiger un PREreview

This article formalizes AI-assisted assessment as a discrete-time algorithm and evaluates it in a digitally transformed higher-education setting. We integrate an agentic retrieval-augmented generation (RAG) feedback engine into a six-iteration dynamic evaluation cycle and model learning with three complementary formulations: (i) a linear-difference update linking next-step gains to feedback quality and the gap-to-target, (ii) a logistic convergence model capturing diminishing returns near ceiling, and (iii) a relative-gain regression quantifying the marginal effect of feedback quality on the fraction of the gap closed per iteration. A case study in a Concurrent Programming course (n=35) shows substantial and equity-relevant improvements: the cohort mean increased from 58.4 to 91.2 (0–100 scale) while dispersion decreased from 9.7 to 5.8 across six iterations; repeated-measures ANOVA (Greenhouse–Geisser corrected) indicated significant within-student change. Parameter estimates further indicate that higher-quality, evidence-grounded feedback is associated with larger next-step gains and faster convergence. We discuss design implications for EdTech at scale (instrumentation, equity-aware metrics, and reproducibility assets) and the relevance of this formalization for comparative analyses of innovative assessment systems. Limitations include the observational, single-course design; future work should test causal variants (e.g., stepped-wedge trials) and cross-domain generalization.