An agentic AI instructor sustained and deepened student engagement across successive course modules in a field deployment¶

Claim · All claims
Evidence · 1 study · qualitative r? · q1
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q1 i? In a graduate Cloud Computing course where an LLM-driven agent served as primary instructor, engagement shifted from broad conceptual exploration in early weeks to deeper, more focused inquiry in later weeks. → Simon Baradziej 2026

Evidence¶

Simon Baradziej 2026¶

Simon Baradziej. (2026). A systematic review of the roles of agentic AI in higher education. Social Sciences & Humanities Open. https://doi.org/10.1016/j.ssaho.2026.103418

q1 · i? · qualitative · r?

Case study by Simmhan and Kulkarni (2025) at the Indian Institute of Science, Bangalore, deploying an AI Instructor Agent within Microsoft Teams Copilot and Moodle as primary instructor, with the human instructor facilitating. Engagement analytics were generated from student-submitted chat transcripts; no effect size is printed.

"The study reported a shift in engagement patterns across successive modules, from broad conceptual exploration in early weeks to deeper, more focused inquiry in later weeks, suggesting that agentic instruction can sustain and deepen student engagement over time."

Discussion¶

Related Claims¶

• Many agentic multi-agent implementations remain at prototype or conceptual stage with limited large-scale deployment and empirical validation — a broader claim this one bears on
• Sustained deployment surfaced four behavioral archetypes — self-directed, tutor-reliant, late engager, and passive consumer — including patterns invisible to the reactive teaching system — related
• The analytics agent surfaced a Module 2 dead zone — students stopping around the 42-minute lecture mark where exercises ended — and the instructor revised the exercise set in response — related
• Early warning and proactive support agents are associated with reported retention and grade improvements in institutional deployments — related
• In an AI-based early warning system, the predictive importance of CAM dimensions shifted over an instructional unit, with affect/metacognition self-reports leading early and cognitive artifacts gaining accuracy toward the end — related
• Current AI deployments in higher education remain task-oriented rather than ecosystem-oriented despite advances in LLMs and multimodal AI — related
• A field study with 10 faculty across two institutions observed a 25% reduction in administrative load and improved LMS data utilization from agentic teaching-assistant systems — related
• A three-week MALA deployment recorded 62 distinct usernames across 128 conversations, with 63% of conversations resolved or partially resolved — related
• In deployment, 13% of sampled LLM agent responses were faulty and 37% very useful, with response quality improving over weeks of use — related