Research

AI Role-Play for Competency-Based Education

This research evaluates generative AI role-play simulations for developing professional competencies in Law and Forensic Science. Using a mixed-methods approach, it validates scalable, AI-mediated practice and assessment frameworks.

Student Help-Seeking Behaviour with AI Tutors

This ongoing research utilises Coursemology data to analyze AI help-seeking in programming. By applying AST-based frameworks, it distinguishes between genuine scaffolded learning and simple task offloading to improve educational outcomes.

Understanding How University Students Really Use LLMs

This ongoing longitudinal NUS study investigates interdisciplinary LLM adoption by comparing self-reported attitudes with actual usage logs. It identifies behavioral gaps to understand how students critically engage with AI over time.

LLM-As-A-Judge Bias

This research addresses LLM-As-A-Judge “agreeableness bias” (TNR < 25%). We propose a minority-veto strategy and regression-based framework using human-annotated data to ensure reliable, scalable, and unbiased evaluation of LLM-generated content.

Codaveri AI-Assisted Feedback

This research at IIT Kanpur explores a hybrid model where Codaveri AI augments Teaching Assistants. This human-in-the-loop approach provides scalable, scaffolded programming feedback, ensuring high-quality instruction and enhanced lab efficiency.