Does 'Be Curious, Not Judgmental' Work on an AI?

We’ve had over 1,000 different people come through a Puerto Rico AI Community meetup at this point, which means I’ve watched a lot of humans meet an AI for the very first time, and I can tell you that they split into two groups within about 90 seconds. It’s not a technical split (some of the sharpest engineers I know land in the wrong group, and some retirees who mostly came for the free coffee land in the right one). Group one asks a question (singular), gets something mediocre back, and files a verdict on the entire field: overhyped and not ready. Then they close the laptop and go tell their coworkers they tried it.

Group two gets the exact same mediocre answer and asks it why.

Ted Lasso comes back Wednesday, which is why all of this is on my mind. You remember the dart scene (everyone remembers the dart scene): Ted’s about to win a bet against a man who’s spent a whole season looking down on him, and he explains that people have judged him his entire life, and not one of them ever got curious enough to ask whether he’d played a lot of darts. Every Sunday, with his father, from age 10. Then Ted pins the lesson on Walt Whitman: “Be curious, not judgmental.”

Whitman never said it. A scholar at the Walt Whitman Archive actually checked, which means a real academic spent part of a real career formally establishing that a poet never said a thing on a TV show. I’d like to buy that man a drink. The quote is still right, even wearing the wrong poet’s name, and I think about it every time I watch somebody in group one walk away from the best research assistant, tutor, editor, brainstormer, travel agent, and patient explainer of tax law that $20 a month has ever bought (and a bunch more jobs I’m forgetting, cause the list grows weekly).

Some of this is just my education talking. I went to Shimer College, a Great Books school (you’ve never heard of it; almost nobody has) where the entire teaching method was sitting around a table asking each other questions (no lectures, no multiple choice, just the Socratic Method applied until you either got smarter or got hungry). For most of my adult life that education was a charming thing to mention at parties and absolutely nothing else. My dad (who had opinions) once asked me what exactly a person does with a degree in asking questions. Took me a couple decades to have an answer for him, and the answer turned out to be “wait for the machines.” The single most useful skill in technology right now is interrogating a confident entity that’s sometimes completely wrong, and it turns out I’d been training for that since before the internet had pictures.

So here’s what the curious group does differently, and you can try this today. When one of my agents hands me a number that’s obviously garbage, I don’t fix it and I don’t file a verdict. I ask it how it got there. And the answer is almost always embarrassing in a useful way: it couldn’t see the actual system, so it did what any of us would do under deadline pressure and made something up that looked plausible. That’s a plumbing problem (mine gets fixed with MCP servers, and I have now explained that acronym something like 400 times in the past year and I remain weirdly delighted every single time, which is either enthusiasm or a condition… my wife has stopped asking which).

My sons will recognize the other half of the pattern. For years, “how was school” got me “fine,” and I accepted “fine” as data (it wasn’t). It was what you get when you ask a closed question with zero context, and the fix was never a better kid, it was a better question. The AI version is identical. “Write me a marketing plan” gets you “fine.” Tell it who you are, what you sell, what happened last quarter, and what almost worked, and suddenly you’re having the other kind of conversation.

The one caveat is real, so I’ll say it plainly (bold claims deserve their caveats): curiosity without verification is how you end up confidently wrong, and I’ve managed that in public more than once. Ted has Coach Beard, who actually reads the books (all of them, apparently). I have a nightstand stack that’s been silently judging me since March. Your AI needs somebody in the Beard chair too, and it’s you… stay curious with the thing, ask it three more questions than feels polite, then check the answer against a source that doesn’t autocomplete.

So, does “be curious, not judgmental” work on an AI?!? Better than it works on people, honestly. The AI never gets tired of you. It won’t sigh, it won’t check its phone, and you can ask the same question 11 different ways at 2 in the morning and it meets you at attempt number 11 with the same patience it had at number one (I have tested this), which is more than I can say for me.

If you’re anywhere near Puerto Rico, come to a PRAIC meetup and watch the two groups form in real time. They’re free (actually free, not conference-free), we meet bi-monthly, and the speakers are great. And if you’re not near the island, write me (Edward@Roske.AI) and tell me about your first conversation with an AI. I promise to be curious about it.

Asking good questions, Edward