Développement d'une plateforme d'apprentissage adaptatif basée sur l'intelligence artificielle

Thesis Abstract

The rapid advancement of educational technologies and artificial intelligence (AI) has significantly transformed the landscape of personalized learning, yet many existing digital learning environments lack the capacity to effectively adapt to individual learner needs in real time. This research addresses the critical challenge of designing and developing an intelligent adaptive learning platform that leverages AI algorithms to tailor educational content and instructional strategies for diverse learners, thereby enhancing engagement and learning outcomes. The primary aim of this study is to develop a robust, scalable platform grounded in AI techniques that dynamically adjusts to learner performance and preferences. The specific objectives include (1) to analyze existing adaptive learning systems and identify their limitations; (2) to design an architectural framework integrating machine learning models for real-time learner modeling; (3) to implement a prototype platform utilizing natural language processing (NLP) and reinforcement learning methodologies; and (4) to empirically evaluate the efficacy of the platform in improving learner engagement, motivation, and achievement within a controlled instructional setting. To achieve these objectives, the study employs a mixed-methods research design, combining quantitative experiments with qualitative user feedback. The population targeted comprises 200 undergraduate students enrolled in an introductory computer science course at a large university, selected through stratified random sampling to ensure demographic representativeness. Data collection instruments include pre- and post-intervention assessments measuring academic performance, engagement scales, and questionnaires evaluating user satisfaction. Additionally, system-generated logs capture real-time interaction data. Quantitative data will be analyzed using statistical techniques such as paired t-tests, ANOVA, and regression analysis to determine the platform’s impact on learning outcomes, while thematic analysis will be applied to qualitative feedback. The anticipated findings suggest that learners engaging with the AI-driven adaptive platform will demonstrate statistically significant improvements in academic achievement and engagement levels compared to traditional educational methods. The platform's ability to personalize content in real-time using reinforcement learning algorithms is expected to increase learner motivation and reduce dropout rates. Furthermore, analyses are expected to reveal correlations between adaptive feedback and sustained learner motivation, supporting the validity of AI-driven personalization in education. This study contributes to the existing body of knowledge by providing a comprehensive model for integrating AI techniques into adaptive learning environments and empirically validating its effectiveness in higher education contexts. It advances understanding of how machine learning, NLP, and reinforcement learning can collaboratively enhance personalized education, and offers a practical framework for implementing scalable adaptive systems. The theoretical foundation draws upon Vygotsky’s Zone of Proximal Development and constructivist learning theories, providing supportive frameworks for the platform's adaptive mechanisms. The main conclusion underscores the potential of AI-based adaptive learning platforms to revolutionize digital education by delivering personalized, engaging, and effective learning experiences. It recommends the broader adoption of such systems in higher education institutions, emphasizing the importance of continuous refinement through learner data analytics. The study also suggests avenues for future research, including longitudinal studies to assess long-term impacts and expansion into diverse educational disciplines. Ultimately, the findings demonstrate that integrating advanced AI techniques into adaptive educational platforms can significantly enhance the quality and accessibility of personalized learning, marking a pivotal step toward intelligent, learner-centered education systems.