Socratic AI Tutor
Guided learning through LLM-powered Socratic questioning
An educational tool that uses large language models not to give answers, but to ask the right questions. Built on the Socratic method, it guides learners...
Research Context: Developed in collaboration with Prof. Léonard Boussioux (UW), Romain Puech (MIT), and Archit Gupta (UW), this project explores how LLMs can be repurposed from answer-generators to pedagogical guides. The research questions whether AI's greatest educational value lies not in explanation, but in structured inquiry.
How It Works: The tutor engages students with contextual hints, counter-questions, and scaffolding prompts. Rather than revealing solutions, it diagnoses misconceptions and nudges learners toward their own discoveries: mirroring the interactive dynamic of human one-on-one tutoring.