Dr. Priyamvada (Pia) Tripathi

I design experiments, cognitive architectures, and deployed AI systems that make agents norm-competent — able to follow, reason about, and legibly deviate from the tacit rules that govern what a good teammate, teacher, or companion may, must, and must not do. My goal is for human trust in AI agents to stay calibrated to what those agents actually deserve.

My research program spans four connected strands. The first asks how an AI agent's norm behavior — adherence, violation, and disclosure — shapes human trust in teams, testing whether a robot's principled deviation can preserve or increase trust rather than erode it. The second develops the technical infrastructure for norm-competent social robots: hybrid symbolic–LLM architectures where a deontic layer holds veto authority over generation, making norm reasoning an explicit, auditable object rather than a statistical tendency absorbed from training data. The third develops a principled framework for designing and validating AI-enabled serious games, grounded in a distinction the field has been conflating: instructional intelligence and instructional adaptivity are separable subsystems with distinct failure modes. The fourth applies this framework to precision medical education — designing AI-based systems for clinical training that close the observe–infer–intervene–measure loop through hybrid assessment architectures that evaluate both what a trainee decided and why they reasoned as they did.

Across all four strands, my methodology combines controlled human-subjects experiments, computational modeling, and design-based deployment in a build–measure–deploy cycle. The unifying commitment is that advancing AI capability and advancing human trust in AI are not the same problem — and that solving the second requires taking the first seriously enough to test it.