Artificial Intelligence and Student Engagement in Online Learning: A Literature Review

Synthesis: Zhou (2025) systematically reviews 24 studies from the Web of Science database on how AI enhances student engagement in online learning. Six key applications emerge: AI chatbots in course design, emotion/facial/voice recognition and eye tracking, machine learning for data analysis, teacher–student interaction support, personalized feedback and recommendations, and AI-powered bots in smart learning environments. Findings show that integrating diverse AI tools and data sources yields more accurate, real-time insight into cognitive, emotional, and behavioral engagement — while limitations include the single-database scope, the conflation of synchronous and asynchronous contexts, and an engagement-only focus.

Key Findings

• Six AI applications support engagement. The review classifies AI use into: (1) AI chatbots integrated into online course design, (2) emotion, facial, voice recognition and eye tracking, (3) machine learning for data analysis, (4) teacher–student interaction support, (5) personalized Feedback and recommendations, and (6) AI-powered bots in smart learning environments.
• Chatbots foster engagement through timely, personalized support. AI-powered chatbots are effective across learning domains and stages — delivering course information, interactive practice, and information retrieval — with comparative analyses finding them especially effective when deployed for personalized, on-demand assistance.
• Affective and behavioral monitoring. Emotion recognition, facial and voice tracking, and eye tracking enable real-time assessment of behavioral, emotional, and cognitive engagement — recognizing states from boredom and confusion to interest and enjoyment using facial expressions, physiological signals, and voice data.(Affective Computing)
• Common data sources and measurement. Studies rely on video recordings, activity logs, standardized datasets, and surveys; engagement is typically measured through multi-method approaches combining surveys, AI recognition, and coded activity data.
• Integration improves insight. Combining diverse AI tools and data sources provides more accurate, real-time insight into students' cognitive, emotional, and behavioral engagement than single-source approaches.

Study Design & Method

A systematic literature review of 24 peer-reviewed studies retrieved from the Web of Science database. The review synthesized how AI has been implemented to foster student engagement in fully online learning (excluding blended and flipped contexts), classifying applications and examining the data resources and measurement approaches used. Engagement is treated as a multi-dimensional construct spanning behavioral participation, emotional investment, and cognitive involvement.

What this means for practice

• Instructional designers. Integrate multiple AI modalities and data sources instead of relying on a single signal — the review's six applications (chatbots, emotion/facial/voice recognition and eye tracking, machine learning analysis, interaction support, personalized feedback, and bots in smart environments) each surface a different facet of engagement.
• Learning analytics designers. Measure Student Engagement across behavioral, emotional, and cognitive dimensions and record which instrument produced each signal, since the reviewed studies draw on video recordings, activity logs, standardized datasets, and surveys.
• Educators. Deploy chatbots for on-demand course information, interactive practice, and support, the mode in which comparative analyses found them most effective — especially relevant in online learning, where self-regulation demands make sustained engagement harder than in person.
• Researchers. Address the open questions the review identifies: real-time interventions, long-term impacts, ethical considerations, and the difference between synchronous and asynchronous online learning.

Limitations

• The review synthesized only 24 studies, selected after a Web of Science search returned 84 records and 59 were excluded as ineligible, with non-English records and items lacking full-text access removed.
• It is a single-database, peer-reviewed-only review; the authors identify both the single-database scope and the engagement-only focus as limitations of what the synthesis can claim.
• The review does not distinguish asynchronous from synchronous online learning, a conflation the authors name explicitly even though design implications differ between the two.
• Studies whose primary focus was not student engagement were excluded, and the heterogeneous engagement measures (surveys, AI recognition, coded activity data) were synthesized narratively, so the review reports no pooled effect estimates.

Citation

Zhou, Z. (2025). Artificial intelligence and student engagement in online learning: a literature review. American Journal of Distance Education. Advance online publication.