What we're doing
In this activity, sub-teams become "experts" on one of the five pillars from OpenAI's blueprint and synthesize their understanding using a Frayer Model — refined through a deliberate cycle of human thinking, AI assistance, and peer collaboration.
Throughout the activity, you'll see two TrAIT designations: AIx for phases where AI is intentionally not used, and AI+ for phases where AI is used to supplement human thinking. The pattern matters — it makes when and why we choose to use AI visible and intentional.
Your synthesis tool
Each sub-team will draft a Frayer Model on poster paper. We've adapted the traditional model to make it more useful for educational leaders — replacing "Characteristics" with "Action Steps for Our District" so the activity ends with practical, actionable thinking.
Find your sub-team
Two sub-teams (a and b) are assigned to each pillar — they'll work independently before comparing in Phase 3.
Three colors, three layers of thinking
Use a different color marker at each phase. By the end, your Frayer Model will be a visual record of how your team's thinking grew — from individual reading, to AI augmentation, to peer collaboration.
Five phases of the AI Sandwich
Watch the TrAIT pattern emerge as you move through the phases. AI enters the workflow exactly once — at Phase 2. Everything before, between, and after is human work.
Open the blueprint PDF on your laptop and read your assigned pillar (page numbers are listed in the group assignments above). Do not open any AI tools — this phase is human thinking only.
Each person takes 3–5 minutes to silently read their assigned section. Then, as a sub-team, collaboratively draft a Frayer Model on poster paper based solely on your own professional knowledge and what you just read.
Don't worry about being comprehensive or "right" — capture your team's authentic first thinking. Some quadrants may have more in them than others. That's the point of Phase 2.
Now bring AI into the conversation. Open Gemini in a new tab. As a sub-team, copy the prompt below and paste it into Gemini, replacing [Pillar Name] with your assigned pillar. For a richer response, also upload the blueprint PDF directly to Gemini along with your prompt.
Review what AI generates against your original Frayer Model. Using your Color 2 marker, add any AI-generated points your team agrees are valuable. Cross out or annotate anything AI got wrong, missed your context, or feels generic. Your Frayer Model should now reflect both human expertise and AI input — with the additions visually distinct.
Close any AI tools. Group 1a meets with Group 1b. Group 2a meets with Group 2b. And so on.
Place your completed Frayer Models side by side. Both teams went through the same process — reading, drafting, prompting AI, and refining — but you likely made different choices.
Walk through each quadrant together and discuss: Did AI give both groups the same output? Where did your teams make different editorial decisions about what to keep? Did one team's human expertise catch something the other team and AI both missed?
Add to your own Frayer Model using your Color 3 marker. You now have three visible layers of thinking on your poster.
Synthesize everything into a final, polished Frayer Model on a fresh sheet of poster paper. You now have three sources: your sub-team's original thinking, AI's contributions, and your partner sub-team's perspective.
You decide what stays, what goes, and what gets rewritten in your own words. AI informed your thinking — it did not replace it.
Return to your home tables. Each table should have at least one representative from each of the five pillars.
Each expert takes 2–3 minutes to teach their pillar to the table using their final Frayer Model. As you listen, ask yourself: How does this pillar connect to what's already happening in our district? What's the gap between where we are and where this blueprint says we should be?
One AI+ phase, surrounded by four AIx phases. The bread is thick. The filling is thin. The human is always in control.
Whole-group reflection
After report-back, we'll come together as a whole group to surface what we learned — both about the blueprint and about the protocol itself.
- Did AI give both sub-teams the same Frayer Model? What does that tell you about AI's consistency — and its limitations?
- Where did human expertise add the most value that AI couldn't?
- Which of OpenAI's five pillars is your district strongest in right now? Which is the biggest gap?
- What did the AI Sandwich process feel like? Could you see your teachers using this same protocol with students?
Use AI responsibly EVERY time
The protocol you just experienced is itself the lesson.
The AI Sandwich works because it builds verification habits into the workflow. Human expertise frames the question, AI expands the thinking, and human judgment decides what stays. That's the same habit your teachers should be building with their students.
Source: https://www.aiforeducation.io/ai-resources/how-to-use-ai-responsibly-every-time (updated Sept. 10, 2024). Click this link for the Long Description of the "How to Use AI Responsibly EVERY Time" image.