What would change your mind?

Here's a question that sounds simple and is actually the whole point of this unit:

"What would change your mind?"

Ask it about anything you believe. If you can answer — if you can name the evidence that would flip you — you're thinking. If you can't name anything, you're not holding a belief. You're holding on.

A real belief comes with an exit. If nothing could change your mind, it was never about the evidence.

One idea: beliefs you can test vs. beliefs you defend

Remember Unit 1: understanding means you could be wrong and notice. This is that idea, turned into a habit.

Kind of statement	Example	What would change your mind?
Testable claim	"This bridge holds 10 tons."	A load test. Clear and checkable.
Opinion	"Blue is the best color."	Nothing — and that's fine, it's a taste.
Sneaky belief	"My team never loses unfairly."	Should be evidence — but often nothing will.

The dangerous one is the third. It dresses up like a fact but you defend it like a taste. The "what would change your mind?" question drags it into the light.

And it's your best defense against a confident AI, too. When a chatbot states something as fact, ask: "What evidence would show this is wrong, and does that evidence exist?" A good answer names real tests. A hallucination can't.

Do the thing

For each statement, answer honestly: what would change your mind? If the answer is "nothing," decide — is that because it's an opinion, or because you're refusing to look?

1. "Our town's water is safe to drink."
2. "Pineapple belongs on pizza."
3. "I'm bad at math."
4. "This AI's answer is correct."

Quick check. #1 and #4 are testable — name the test (a water report; an independent source). #2 is pure opinion — nothing changes it, and that's honest. #3 is the sneaky one: it feels like a fact about you, but "I'm bad at math" is a prediction that a few good weeks could break. Beliefs about yourself are the hardest to put an exit on — and the most worth it.

Why this matters

You finished Unit 4 — the most important unit in the course. Here's what you carry out:

• Confidence is not evidence (from anyone — an AI, a friend, or you).
• Check the load-bearing claim, in a source you can trace.
• Ask "what would change my mind?" — and mean it.

That last question is what keeps you the boss of your own thinking in a world full of confident answers. It's the difference between being convinced and being correct.

Now you're ready to build. Next unit, you'll take everything — the verbs, the data sense, the prompting, the checking — and design a tiny AI workflow that actually does something useful.