I saw a video a while ago that I haven't been able to stop thinking about.
A girl turns up to an Ariana Grande concert and finds out she can't get into the venue with the bag she's carrying. Clear plastic only. Security won't budge. She doesn't have one, nobody outside is selling one, and neither does anyone else in the queue.
So she goes home and orders a hundred clear plastic bags off Amazon. Comes back the next night, sets up outside the same venue, and sells them for five dollars each. Sells out. Goes back, restocks, sells out again. Somewhere around a thousand dollars in two nights.
There's no AI in that story. That's actually the point.
The skill underneath
What she did was notice a problem while standing in the middle of it.
Everyone in that queue had the same problem. Most of them were irritated about it, sorted it out for themselves, and forgot about it by the time the lights went down. She was the one who looked at it and thought: this isn't only my problem. This is four thousand people's problem, and the answer costs eleven cents.
That's it. That's the whole skill. Notice the problem, work out how many people have it, solve it. It's the foundation under every business anyone has ever built and it has nothing to do with technology at all.
And it's the thing I most want to teach my kids.
Not because I'm hoping to raise founders. I'm really not. It's more that this way of looking at the world seems useful no matter what they end up doing. A person who can see a problem clearly and think their way to an answer is going to be okay in most rooms they walk into.
The world they're actually growing up in
Here's the part that sits underneath all of this for me.
My two are one and three. Anyone born from about 2025 onward is growing up in a world where AI was already here when they arrived. There's no before for them. It's not a new thing that turned up and had to be argued about and got written up in the news. It's just part of the furniture, the way plumbing is.
I grew up in the nineties. I watched the internet arrive. I remember the sound of it connecting. My kids will never have that experience with any of this, and I think that changes things in ways I can't fully see yet.
Which makes the obvious instinct to teach them the tool. Show them how to prompt, how to get good output, how to work with the thing.
But I don't think that's it. Whatever my three year old ends up using in 2045 doesn't have a name yet. Teaching her today's tool is like teaching someone to use a specific fax machine. The interface is the least durable part of any of this.
The durable part is knowing what you want from it. And knowing what you want from it starts with being able to see a problem clearly in the first place.
So the sequence I keep coming back to is small and boring. Notice something that's annoying. Ask whether it's annoying for other people too. Work out what would fix it. And then, only then, ask whether AI has anything to do with the fix.
The question I'm still learning to ask
That last step is the one I'm worst at.
I reach for AI constantly. Sometimes for things that would genuinely take me four minutes by hand. I've spent an hour building something clever to avoid a task that was ten minutes of tedium, and I've felt smart the entire time I was doing it.
So the thing I want to teach my kids is a thing I'm still teaching myself. It's a decision tree, and there's a branch on it I keep skipping.
Here's a problem. I think I know how to solve it. Do I actually need AI for this, yes or no?
Sometimes the answer is no and the honest reason is that using it would be more fun. Sometimes the answer is yes for one small piece of the job and no for the rest of it, and being able to tell the difference is most of the skill.
That question is where the critical thinking lives. Not in the using. In the deciding whether to use.
And that's the difference between being the person directing this thing and being a person the thing directs. I want my kids on the directing side of that line. I want them creative enough to decide what should exist, and clear-headed enough to know when the tool has nothing to offer.
They're one and three, so none of this is a lesson yet. Right now it's just noticing things out loud with them.
But what I want them to end up with is simple enough to say in one breath. Find the problem. Work out how you'd solve it. Then decide whether you actually need AI for it. If the answer is no, that's a good answer. If it's yes, we'll figure that part out together.
That girl in the queue didn't have a framework. She had a problem and she paid attention to it. Everything else is just whichever tool you happen to pick up afterwards.