Explainable Artificial Intelligence (XAI) and Human-in-the-Loop (HITL): Technology and Teacher Mediation in Formative Feedback

Abstract

Artificial intelligence (AI) has been incorporated into education as a resource capable of optimizing processes and personalizing learning, but its use raises challenges related to explainability, fairness, and pedagogical mediation. This study aimed to analyze the integration of Explainable Artificial Intelligence (XAI) and the Human-in-the-Loop (HITL) model in providing formative feedback in digital environments. To this end, we adopted a theoretical-reflective approach based on an interpretative literature review, aiming to problematize relevant academic productions and systematize contributions on the topic. The results indicate that, in isolation, AI presents limitations such as the reproduction of biases and the lack of pedagogical contextualization; however, when combined with teacher mediation in hybrid protocols, it enables the production of clear, personalized, and inclusive feedback. We conclude that the articulation between XAI and HITL represents a promising path to strengthening flexible learning and ensuring more ethical, transparent, and equitable assessment practices, in which technology complements, but does not replace, pedagogical action.