Misconception Hunter

Misconception Hunter is a voice-native AI tutor, built on Agora Conversational AI, that investigates how a student reasons about a CS/AI concept instead of just grading their final answer. Its goal is to catch misconceptions that a normal Q&A tutor would miss — a student getting the right answer for the wrong reason, or a real misunderstanding hiding behind a confident tone — by asking Socratic follow-up questions, tracking contradictions across the session, and only naming a misconception once it has real evidence from multiple turns, never from a single wrong answer. At the end of each session it produces a structured report (topic, misconceptions found with evidence, strengths, and a human-escalation flag) so a teacher can act on it without reviewing the full transcript.