The Science Fair Model

A lot of organizations feel like they need a comprehensive strategy before they can start with AI. A committee, a policy document, board approval, a vendor evaluation. That’s understandable — these are careful organizations, and careful is good.

But there’s a lighter way to start. The Science Fair Model is a one-week experiment that keeps the stakes low while teaching your team more about AI than any strategy document could.

Before we get into the mechanics, here’s the bigger picture: one of the clearest trends in AI is that it’s making it easier and easier for people to build tools, apps, and technology. Within months, it will become increasingly common for someone on your team to build custom tools for the team in-house — things you used to have to buy as software. The Science Fair is the precursor to that culture of building and experimenting. It’s not just about learning to use AI. It’s about building the organizational muscle for a world where your team creates its own solutions.

The Science Fair isn’t just about learning to use AI. It’s the precursor to a culture where your team builds its own solutions.

Before You Start

A few things to get right up front:

Budget for a paid tool. Give your team access to a paid AI chatbot — Claude, ChatGPT, Gemini, whatever you prefer. Paid tiers have better data handling, as we covered in the data privacy lesson. This is a small investment that practices good habits from day one.

Follow the MVP Policy fundamentals. The Science Fair isn’t a free-for-all. Be thoughtful with sensitive data. Use tools from reputable providers. Apply review to anything that’s going external. The fundamentals from Module 3 should guide the experiment.

Make it fun. This works best when it feels like exploration, not compliance training. The energy should be curiosity, not obligation.

How It Works

Monday: Gather problems. Ask every team member to submit one or two tasks that eat their time. Not AI-specific tasks — just frustrating, repetitive, time-consuming work. “I spend four hours every month reformatting budget narratives.” “I rewrite our mission statement slightly differently for every application.” “I manually search three databases to find new funders.”

Collect these on sticky notes, a shared doc, or a whiteboard. Don’t filter yet.

Tuesday through Thursday: Experiment. Each person picks one of the problems — their own or someone else’s — and tries using AI to solve it. No pressure to succeed. No formal process. Just: “Can AI help with this? Let’s find out.”

Give your team explicit permission to spend 30-60 minutes per day on this. It’s not a distraction from work — it’s an investment in capacity.

Friday: Present. Like a science fair, each person shows what they tried. What worked? What didn’t? What surprised them? What would they try differently?

Why It Works

Removes the pressure of “getting it right.” This is an experiment. Experiments are allowed to fail. That permission is what makes people willing to try.

Starts with real problems. You’re not asking “what can AI do?” You’re asking “can AI help with this specific thing that already frustrates us?” That’s a more productive question.

Creates shared experience. After one week, your entire team has firsthand experience with AI. They’ve seen what works, what doesn’t, and where the surprises are. That shared understanding is worth more than any training session.

Produces immediate value. Even in one week, teams typically find two or three genuinely useful applications. “I can draft a first pass of the LOI in 20 minutes instead of two hours” is a real, immediate win.

Philip’s Take: Make it fun. Treat it like a puzzle. Everyone brings a real frustration, and the challenge is: can AI help solve it? There’s a decent chance somebody’s going to figure out something that makes everyone’s workflow a lot better — and the whole team learns about AI along the way. If people enjoy it, make it a regular thing. That experimentation muscle is what builds the culture for a future where your team isn’t just using tools — they’re building them.

Running Your Own Science Fair

Practical tips:

Pro tip

Make sure people have access to AI tools before Monday. If they have to request accounts and wait for IT, the momentum dies. Have paid accounts set up and ready.
Frame it as an invitation. “Let’s spend this week exploring something together.” Voluntary energy beats mandatory compliance.
Protect the Friday time. Don’t let the presentations get bumped. They’re the payoff.
Document what people find. This becomes your first set of shared learnings (Component 4 of the MVP Policy).
Celebrate failures. When someone says “I tried using AI for X and it was terrible,” that’s just as valuable as a win. Now everyone knows not to waste time on that.

After the Science Fair

If the experiment goes well, you have natural next steps:

The useful discoveries become part of your workflow
The MVP Policy fundamentals (Module 3) give structure to ongoing AI use
The shared learnings become your first entries in the team knowledge base
The momentum carries into deeper adoption — and eventually into building custom tools

If the experiment doesn’t generate much excitement, that’s useful information too. You’ve learned where your team stands, and you can try again in a few months when the tools have gotten even better.

Key Takeaways

The Science Fair Model is a low-stakes, one-week way to introduce AI to your team
Budget for a paid tool and follow the MVP Policy fundamentals -- practice good habits from day one
Monday: gather real problems. Tuesday-Thursday: experiment. Friday: present results
This is the precursor to a culture of building -- where your team eventually creates its own tools and solutions

### Next Lesson

The Science Fair gets you started. But how do you expand from there? The Low-to-High Risk Ladder gives your team a progression from comfortable experiments to more complex AI use.