AI-Enhanced Pedagogy: A Critical Review of Intelligent Systems for Fostering Multidimensional Student Engagement.

This structured critical review examines how artificial intelligence (AI) systems foster multidimensional student engagement—behavioral, cognitive, emotional, social, and agentic—in contemporary learning environments. Using a PRISMA-informed process, we searched Scopus, Web of Science, ERIC, IEEE Xplore, and ACM Digital Library for diverse English-language studies (2020–2025). Eighty-eight records were identified; 48 were screened for descriptive trends and 26 were synthesized. We map major AI classes (intelligent tutoring systems, learning analytics, generative AI, affective computing) to engagement mechanisms and associated features (adaptivity, feedback, dialogue agents, dashboards, affect sensing). Evidence shows AI most consistently strengthens cognitive and behavioral engagement through guidance, personalization, and analytics-driven support, while emotional engagement remains underdeveloped and inconsistently evaluated. We argue that generative AI can scale agentic engagement via learner-initiated prompting, co-creation, and self-regulation, but amplifies risks including hallucination, overreliance, academic integrity, privacy, and bias. We conclude with a human-in-the-loop framework and systemic AI literacy to align innovation with ethical governance.