Ep09 - Resilience Over Right Answers: Rethinking Science Education in the Age of AI (with Biophysicist Jon Sack, UC Davis)

Science education after ChatGPT: what happens when students can outsource the thinking, and still turn in something that looks right? In this episode of My Robot Teacher, CSU professors Sarah Senk and Taiyo Inoue talk with UC Davis biophysicist Jon Sack about AI literacy, scientific thinking, and how LLMs are reshaping both the classroom and the day-to-day reality of research.If the most available “mentor” in a student’s life is an LLM optimized to validate, what happens to the virtues science depends on: tolerating disconfirmation, staying curious through failure, and separating confidence from evidence? And if AI can generate the output, what exactly are we teaching - especially when the point is conceptual understanding, not polished answers?In this conversation, we explore:What AI literacy should mean in science classrooms (beyond “don’t cheat”)How to resist the reward of feeling right when LLMs produce fluent, plausible explanations on demandHow to redesign assessment so students can’t simply outsource the thinkingWhat “good” use looks like: prompting for falsification instead of praise, plus habits of verification and iterationWhat AlphaFold and protein design teach us about hypothesis overload, “hallucinations,” and selection under uncertaintyThe bigger meta-question: if we’re co-evolving with AI, how do we keep student agency intact?Ultimately, Jon argues that resilience isn’t a soft skill in science—it’s the method: reality-testing what sounds plausible (including AI-generated ideas) and iterating without outsourcing the thinking.Sponsored by the California Education Learning Lab.💬 Drop your perspective in the comments. We may feature listener takes in a future episode.✅ Subscribe for more “in the wild” classroom experiments and AI literacy for educators.CHAPTERS00:00-6:27 - Chapter 1 - Introduction: Claude Code Built My Canvas Course (Winter Break Experiment) 6:28-10:14 - Chapter 2 - Jon Sack’s First ChatGPT Moment (and the “Too-Positive” AI problem)10:15-12:44 - Chapter 3 - Resilience is the Core Skill in Science12:45-16:08 - Chapter 4 - Scientific Method = Falsification: “Kill Your Darlings” and Reality Testing16:09-19:26 - Chapter 5 - Conceptual Understanding vs. Outsourcing: When the Thinking is the Assignment19:27-25:25 - Chapter 6 - AL Literacy for Students: Use Every Tool, Track Limits25:26-28:22 - Chapter 7 - Inside Jon Sack’s Lab: Ion Channels and Stochastic Decisions28:23-30:36 - Chapter 8 - Stochastic 101: Probability, Sampling, and Why LLMs Vary30:37-38:55 - Chapter 9 - Are We Stochastic All the Way Down? 38:56-44:07 - Chapter 10 - AlphaFold & Protein Design: Cheap Hypotheses, Hallucinations, Verification44:08-53:30 - Chapter 11 - Co-evolving with AI: are tools optimizing around us, and are we changing around them?53:31-1:01:22 Chapter 12 - Education After ChatGPT: Epistemic Virtues, Judgment, and Student Agency