As AI systems generate fluent answers instantly, traditional assessments struggle to distinguish performance from understanding. This session introduces a judgment-centered teaching framework that uses AI as productive friction rather than a shortcut. Participants will explore how deliberately designed prompts, contradictions, and AI-generated confidence can expose reasoning, surface misconceptions, and support deeper learning without relying on surveillance or detection tools. Practical classroom examples from government, history, and composition courses will illustrate how authority can be exercised through question design rather than answer control.
