Georgia Tech's Ivy AI Coach now diagnoses learner misconceptions with 2-model system

Intelligent tutoring systems have long excelled at generating explanations but rarely provide principled diagnosis of where and why a learner is wrong. Researchers at Georgia Tech address this gap by introducing a misstep-aware coaching capability for Ivy, their neurosymbolic AI coach used in an online graduate AI course. The system employs a two-model architecture that augments the existing Task-Method-Knowledge (TMK) model with a new Pedagogical Model (PM). The PM makes instructor diagnostic knowledge explicit and machine-readable by encoding, for each quiz question and incorrect response, the learner's underlying belief (a brief statement of the incorrect idea or missing knowledge), a TMK locus (the source of the misunderstanding), a misconception type, and targeted scaffolding derived from the instructor's Q&A key.

Using quiz questions from the course, the team demonstrates a proof-of-concept pipeline that detects and classifies learner errors and generates diagnosis-grounded scaffolding. This moves Ivy beyond knowledge retrieval toward diagnostic misstep awareness, enabling more precise, actionable feedback that supports conceptual change. The work represents a significant advance for adaptive learning systems in AI in education and the learning sciences, shifting the paradigm from 'what to explain' to 'what to diagnose.' By formalizing instructor diagnostic knowledge into a reusable model, this approach could scale personalized error analysis across courses and institutions, potentially transforming how AI tutors interact with students.