Higher‑education experiments with AI instructors moved into the spotlight as Georgia Tech and edX pilots deployed avatar‑based, AI‑driven lecture delivery. Georgia Tech’s David Joyner and MIT’s Anant Agarwal are testing a virtual instructor—an avatar dubbed DAI‑vid—to scale generative‑AI coursework, part of broader efforts to meet rising demand for AI training and workforce upskilling. At the same time, researchers warn about AI behavior that complicates classroom use: a recent paper from OpenAI researchers highlighted that models are trained to avoid uncertainty language like “I don’t know,” which can produce confident but incorrect responses. Educators and instructional designers are confronting trade‑offs between scalability, assessment integrity and modeling epistemic humility for students learning to evaluate information and exercise judgment with AI tools.