Foundation Models and Generative AI
MIT OpenCourseWare offers this non-technical lecture series on the ideas behind ChatGPT, Copilot, CLIP, Dall-E, Stable Diffusion, and AlphaFold. The course opens with a short history of artificial intelligence, then explains what supervised learning and reinforcement learning leave out, building toward self-supervised learning and the foundation models it produces. Lectures cover how these models are trained and applied across both scientific research and business settings, including debates over whether current systems approach artificial general intelligence. No coding or heavy math is required, and the material is pitched at listeners from any background who want to understand the mechanics and implications of generative AI rather than build it themselves. Materials consist of recorded lectures following MIT's standard OpenCourseWare format, free to access with no certificate offered.