This thesis investigates the enrichment of Technology-Enhanced Learning Environments (AIED), especially Intelligent Tutoring Systems (ITS), through the integration of mechanisms that foster self-regulation—a key determinant of academic success. Building on Zimmerman’s socio-cognitive model, it focuses on two core dimensions: self-assessment and self-efficacy beliefs. The study relies on a large-scale reading-instruction ITS deployed in primary schools.
An experimental layer was designed and tested with thousands of primary school pupils, combining teacher surveys, interaction-trace collection, and machine-learning methods. The findings show that young learners frequently struggle with self-assessment and that their confidence in their abilities remains fragile. However, the introduction of metacognitive prompts, tailored feedback, and enriched interfaces helps prevent or correct these shortcomings, while improving persistence and motivation.
The thesis proposes a methodological framework for integrating self-assessment and self-efficacy support modules into AIED systems, demonstrating their effectiveness and scalability. It also opens perspectives for the use of artificial intelligence to provide contextualized support, the development of shared teacher-student dashboards, and the extension of experiments to other educational levels and contexts.