Planned Agenda — Workshop 2: Verifying STEM AI Outputs

8:30

Opening15 min

Icebreaker + Pre-Survey

Drop in chat: Have you ever been fooled by something AI generated? One sentence — bonus points for a funny story. Then complete the pre-survey while we get settled.

→ Open Pre-Survey

8:45

You DoContextualize15 min

CRAFT Orientation + Self-Assessment

Quick CRAFT cycle intro (C→R→A→F→T) and the stakes: your students are already using AI. In the shared doc, answer honestly: How are your students using AI right now? (Not at all / Secretly / With my guidance / I have no idea.) Then: What's your biggest worry about student AI use?

→ Open Shared Doc

9:00

ListenReframe15 min

Poll + The Real Problem Isn't AI Use

When a student uses AI: (a) plagiarism, (b) they didn't learn, (c) it might be wrong, (d) I can't tell. Vote honestly. Then the key reframe: banning AI teaches avoidance; teaching verification teaches engineering thinking. We don't ban calculators — we teach when and how to use them.

9:15

You DoReframe15 min

Breakout: Rewriting the AI Policy

Your group is an AI policy committee. Draft 3 classroom norms for AI use that assume students WILL use AI but must VERIFY outputs. Think: what does "acceptable AI use" look like in your subject? Write your norms in the shared doc.

9:30

Break15 min

Break #1

Check chat for a deliberately wrong AI "fun fact." Can you spot the error before we're back?

9:45

You DoAssemble30 min

I Do + Breakout: Code Error Hunt

First, watch us debug a seemingly-correct AI-generated Python function that's actually wrong. Then in your breakout group: you get 3 AI-generated code snippets (math/science). For each — Read it. Predict what it does. Find the error. Fix it. Document your findings in the shared doc. First group to catch all 3 wins bragging rights.

→ Open Code Snippets

10:15

You DoAssemble15 min

Solo: Break Your Own AI

Ask any LLM to solve a code or math problem from your subject area. Try to break it. Document one error you found (or if it's correct, explain exactly how you verified it). Add to the shared doc.

10:30

You DoAssemble30 min

I Do + Breakout: Content Error Hunt

First, watch an LLM explain a STEM concept — confident, correct-sounding vocabulary, subtly and dangerously wrong. Then in your group: audit AI-generated science/math explanations using the Content Audit Checklist. Check claims, verify facts, assess grade-appropriateness, spot misleading simplifications. Flag the sneakiest error for share-out.

→ Open Audit Checklist

11:00

Break15 min

Break #2

Preview the Check the Machine one-pager — we're about to go deep on this.

→ CtM One-Pager

11:15

You DoFortify30 min

Check the Machine: Live Cycle, Partner Practice & Build Your Own

The 4-step protocol: Task → Before → After → Takeaway. We run a full cycle on screen — follow along on your device. Then pair up to run your own CtM on a prompt from your subject (partner asks: "Did you actually verify, or just skim?"). Finally, customize the editable CtM template for your students: grade-level language, sample task from your subject, student-facing instructions. This is yours to take home.

→ CtM Template

11:45

Transfer + CloseTransfer15 min

CRAFT Debrief, Pair-Share Commitment & Post-Survey

Quick debrief — every activity today mapped to C→R→A→F→T (you just lived two nested CRAFT loops: code and content). With a partner: What's one concrete step you'll take to build a verification-first culture this semester? "I will..." not "I might..." Then complete the post-survey. Your students don't need to be AI experts — they need to be AI skeptics. That's a skill you just learned to teach.

→ Open Post-Survey

Verifying STEM AI Outputs: What You Really Need to Know