Object structure

learning analytics ; AI tutoring systems ; undergraduate Physics ; personalized learning ; educational equity ; actor-network theory (ANT) ; technology acceptance model (TAM) ; Bourdieu's theory of practice ; digital pedagogy ; data-driven education

This article presents a comprehensive analysis of data-driven learning technologies—specifically Learning Analytics (LA) dashboards and Artificial Intelligence (AI) tutoring systems—in undergraduate physics education. Emphasizing the importance of pedagogical integration and sociological context, the study explores how these tools influence learning outcomes, student engagement, and equity. Learning Analytics dashboards are shown to support engagement and self-regulation, particularly when integrated with active learning pedagogies. However, their direct impact on academic achievement remains inconsistent, with effectiveness hinging on design and implementation. AI tutoring systems, including cognitive, dialogue-based, and generative models (such as RAG-based LLMs), display greater promise in enhancing conceptual understanding, problem-solving skills, and personalization. Their success depends not only on technological capability but also on the alignment with learner needs, faculty acceptance, and social equity. The study employs a triangulated methodology combining international case studies, sociological theory (TAM, ANT, Bourdieu), and synthesized survey data to assess user perceptions. It identifies key barriers, such as technological fluency gaps, digital divides, and ethical concerns around data privacy, algorithmic bias, and over-reliance on automation. A focused lens on Armenia’s context underscores infrastructural and pedagogical challenges limiting adoption. The article concludes with a critical synthesis: data-driven tools can significantly enhance physics education but are not panaceas. Their success depends on context-sensitive pedagogical integration, faculty and student readiness, and ethical design. Recommendations emphasize hybrid human-AI models, explainable AI, and equity-first deployment strategies.

Երևան

Զանգակ հրատ.

Հոդված

pdf