The beauty of the glitch: Why AI imperfection is the future of learning

Generative artificial intelligence (GAI) has arrived in the classroom! Yet, educational discourse remains trapped in a defensive crouch, obsessing over how to “sanitize” GAI outputs or mitigate its imperfections (e.g., hallucinations, biases). In this ignite talk, I will argue that the quest for AI accuracy within education is a pedagogical dead end. Instead, I will introduce novel instructional designs that intentionally harness the “glitch” – leveraging AI’s misleading outputs, creative leaps, and logical failures as raw material to fuel human learning and assessment. By trading “perfect” AI for “productive" AI, we can transform a tool of convenience into a catalyst for friction, creating the need for learners to think and collaborate critically while mastering the messy, quintessentially human art of navigating an uncertain world.