Joanne Mumley jumped on AI as soon as it opened up, mostly to stay in step with her middle schoolers. What she found changed how she thinks about assessment. If an AI system can produce the final product from a single prompt, the product no longer proves the learning happened.

In this episode, Joanne explains how her district curriculum team stress-tests performance tasks for AI durability, running old assignments through AI systems to see what survives, then rebuilding the ones that fail so the productive struggle stays with the student. She describes why she opens the school year by asking students how to make good decisions in a world where information is easy to create and difficult to evaluate. And she tells the story of Jimothy, a short-spined raccoon from Seattle’s Ballard neighborhood whose real-or-fake Reddit debate became her first source evaluation lesson of the fall.

We get into her state-level run in the Presidential AI Challenge, where judges asked her to defend why she built with Claude instead of Gemini, and the painting the internet trashed as AI slop. It was a real Monet.

Joanne closes with advice for department heads tempted to write policy before they understand the tools. And she names what she hopes Wolcott graduates carry out the door: the habit of looking for evidence before forming an opinion, and the freedom to change their minds when the evidence changes.

Joanne Mumley is the ELA Curriculum Coordinator and High School English Department Head for Wolcott Public Schools. With over 20 years of classroom experience, she works with educators to design curriculum and develop AI-literate classrooms. Joanne is the 2026 Connecticut State Winner of the Presidential AI Challenge and has presented at state and national conferences, including NEATE and CTCTE, on AI literacy, curriculum design, and assessment in the age of artificial intelligence.