The BEAM (Bot-Enhanced Assessment for Meta Learning) model was introduced in GEN2016 Science and Better Life to reimagine the assessment process and actively engage students in their own learning. Initially piloted with a cohort of 40 students during the Summer Semester of AY2024/25 and later scaled to serve over 400 students, this innovative model fostered remarkable participation, with students generating more than 20,000 chatbot exchanges across three semesters.
A key feature of this initiative was the deployment of topic-specific AI tutors, each aligned with the course’s scientific themes and served as formative partners in assessment, providing students with real-time, personalized feedback as they developed their work. Students were empowered to co-design assessment rubrics, customize their AI tutor interactions according to their project goals, and reflect continuously on their learning journey. By taking an active role in setting marking criteria, students gained a clearer understanding of assessment standards, appreciating the importance of judgment, realism, and accountability in academic work.
Students were encouraged to engage in sustained, multi-turn dialogues with the AI tutors, seeking clarification, probing deeper, and revisiting their discussions as their projects progressed. This iterative feedback cycle supported students in identifying areas for growth, refining their submissions, and making tangible progress, as evidenced by the improved scores and more insightful comments.
These experiences demonstrated that AI tutors can play a powerful role in supporting self-regulated learning. Rather than viewing assessment as a final checkpoint, students engaged in an ongoing, reflective dialogue that bridged the gap between initial understanding and their learning goals. The BEAM model showcased the potential of technology-enhanced formative assessment to inspire participation, reflection, and transferable skills, making the process of learning both dynamic and student-centered.
