Transforming classroom conversations: Cornell University’s AI-powered Socratic Chat on AWS

This article describes how Cornell University developed Socratic Chat, an AI-powered teaching assistant built on AWS that delivers personalized Socratic instruction at scale.

• AI assistant guides students through progressively harder questions based on Bloom's Taxonomy
• Over 500 students across multiple courses used the tool in fall 2024
• Built using Claude by Anthropic in Amazon Bedrock with Streamlit and FastAPI frameworks
• Integrated directly into Canvas learning management system for easy adoption
• Mirrors instructor's teaching style by training on lecture transcripts
• Students spend 10-15 minutes per assignment working through adaptive learning levels
• Automatically generates quiz questions to identify student knowledge gaps
• Frees professors for in-depth instruction and face-to-face interactions
• Expanding to additional courses and collaborating with other institutions

Cornell's approach demonstrates how AI can scale personalized instruction while maintaining human connection and learning integrity in higher education.