JIA Webapp

NSF National AI Institute for Student-AI Teaming

At iSAT, we are starting a new initiative to train and build an LLM-powered experience for students and educators. We are creating an agent that will assist students with typical worksheet problems autonomously, reducing load on teachers and educators while achieving education goals more often than current solutions. We aim to use LLM-powered agents to encourage students to arrive at conclusions themselves, and intend to use them to keep student work directed, effective and on-topic.

The JIA Webapp is the spear of this solution, and is currently being architected, written, designed, and developed by me with support from John Weatherly on the architecture, the iSAT team is in charge of agent R&D, and Elsy Meis for UI/UX work. I’m working on this as part of the SHINE Lab’s team, led by Dr. Leanne Hirshfield.

Photos

JIA student worksheet beside an AI chat panel, with a confirm dialog before submitting an answer — Early Prototype of One-Question-At-A-Time with Confirm Dialog

JIA student worksheet with a numbered question stepper and an open chat with the AI partner — Early Prototype of One-Question-At-A-Time

Publication

Emily Doherty, E. Margaret Perkoff, Sean von Bayern, Rui Zhang, Indrani Dey, Michal Bodzianowski, Sadhana Puntambekar, Leanne Hirshfield. 2025. Piecing Together Teamwork: A Responsible Approach to an LLM-based Educational Jigsaw Agent. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, 1–17. ACM. doi:10.1145/3706598.3713349

I was responsible for architecting and building out the web platform for teachers and students, allowing students to collaborate live with an LLM powered agent under educator supervision.