IJIET 2026 Vol.16(8): 2233-2245
doi: 10.18178/ijiet.2026.16.8.2682
doi: 10.18178/ijiet.2026.16.8.2682
Beyond Generic Feedback: Personalized Generative-AI to Support Calibration and Performance
Hassna Akhasbi 1,*, Najib El Kamoun 1, Fatima Lakrami 1, Jean-Luc Gilles 2,
Jean Michel Rigo 3, and
Emilio Aliss Paredes 4
1. Department of Physics and Engineering, Faculty of Science, Chouaib Doukkali University, El Jadida, Morocco
2. Training and Research Unit in Teaching, Learning and Assessment, University of Teacher Education, Haute École Pédagogique du Canton de Vaud, Lausanne, Switzerland
3. Department Capacity Group Neurosciences, Faculty of Medicine and Life Sciences, Hasselt University, Hasselt, Belgium
4. Department Coordinator of the Center for Academic Development and External Postgraduate Studies (CEDAPE), Faculty of Science and Technology, Universidad Simón I, Patiño, Bolivia
Email: hassna.akhasbi@gmail.com (H.A.); elkamoun.n@ucd.ac.ma (N.E.K.); lakrami.f@ucd.ac.ma (F.L.); jean-luc.gilles@hepl.ch (J.-L.G.); jeanmichel.rigo@uhasselt.be (J.M.R.); emilioaliss@gmail.com (E.A.P.)
*Corresponding author
2. Training and Research Unit in Teaching, Learning and Assessment, University of Teacher Education, Haute École Pédagogique du Canton de Vaud, Lausanne, Switzerland
3. Department Capacity Group Neurosciences, Faculty of Medicine and Life Sciences, Hasselt University, Hasselt, Belgium
4. Department Coordinator of the Center for Academic Development and External Postgraduate Studies (CEDAPE), Faculty of Science and Technology, Universidad Simón I, Patiño, Bolivia
Email: hassna.akhasbi@gmail.com (H.A.); elkamoun.n@ucd.ac.ma (N.E.K.); lakrami.f@ucd.ac.ma (F.L.); jean-luc.gilles@hepl.ch (J.-L.G.); jeanmichel.rigo@uhasselt.be (J.M.R.); emilioaliss@gmail.com (E.A.P.)
*Corresponding author
Manuscript received December 31, 2025; revised February 19, 2026; accepted April 7, 2026; published August 21, 2026
Abstract—Generative Artificial Intelligence (GenAI) opens new possibilities for formative assessment, yet most existing systems still provide generic, task-focused comments that say little about how students manage uncertainty. This exploratory pilot study examined the feasibility of an indicator-informed GenAI workflow that combines Degree of Certainty (DC)-based assessment with course-grounded feedback in a Master’s-level Artificial Intelligence course. Over one semester, 23 students completed five formative DC-enhanced multiple-choice assessments and one summative assessment. A Python-based engine computed cognitive performance (Percentage of Correct Responses (TRC)) and four calibration indicators: Realism/Calibration (REAL), Overestimation (SUR_EST), Underestimation (SOU_EST), and Net Bias (CENTR). These indicators, together with item-level responses and instructor-approved course documents, were fed into a GPT-4 module through a structured prompt and a canonical JavaScript Object Notation (JSON) template specifying six feedback cases (correct/incorrect × low/medium/high certainty). For each assessment, the system generated item-level explanations and a metacognitive synthesis report, which the instructor reviewed before release. Students also completed a perception questionnaire covering usefulness, clarity, trust, intention to use, cognitive load, and self-efficacy. Results showed higher TRC and REAL, lower SUR_EST and CENTR, and stable SOU_EST within an a priori equivalence margin. Perceptions were neutral to slightly positive, with reliable subscales and associations suggesting that perceived clarity and intention to use were linked to more favorable trajectories in performance and calibration. Given the small sample, single-course context, and absence of a control group, these findings remain hypothesis-generating. They nonetheless support the feasibility of indicator-informed, course-grounded GenAI feedback for fostering more realistic confidence judgments and self-regulated learning in higher education.
Keywords—Generative Artificial Intelligence (GenAI), Large Language Models (LLMs), Degree of Certainty (DC), metacognitive calibration, formative feedback, higher education
[Supplementary]
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).
Keywords—Generative Artificial Intelligence (GenAI), Large Language Models (LLMs), Degree of Certainty (DC), metacognitive calibration, formative feedback, higher education
[Supplementary]
Cite: Hassna Akhasbi, Najib El Kamoun, Fatima Lakrami, Jean-Luc Gilles, Jean Michel Rigo, and Emilio Aliss Paredes, "Beyond Generic Feedback: Personalized Generative-AI to Support Calibration and Performance," International Journal of Information and Education Technology, vol. 16, no. 8, pp. 2233-2245, 2026.
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).