Veritasium: What Everyone Gets Wrong About AI and Learning – Derek Muller Explains

Caption notes

This Perimeter Institute talk uses past promises of educational “revolutions” to argue that new technologies tend to become tools in teachers’ hands, while learning depends on repeated, effortful practice and feedback. Derek Muller draws on cognitive load, chunking, and expert–novice differences to argue that skilled performance depends heavily on experience in the relevant domain. He favors worked examples, gradually withdrawn support, and practice toward mastery. For AI, he sees promise in timely feedback and practice support, while worrying that generative tools can replace necessary work such as writing. The Q&A extends these themes to assessment, teacher relationships, scale, and AI-assisted learning.