Over the weekend I was checking my son’s writing assignment. The entire essay could be summed up in one sentence – the play date was fun because they played video games.

Fun how? With whom? Which game? What moment? Nothing. Just fun.

I asked one question: where exactly does it show that it was “interesting”? He thought for a moment, picked up the iPad, and opened ChatGPT: “How do I write an interesting story?”

ChatGPT gave him a perfect framework – start with a hook, build conflict in the middle, end with a reflection. Then what? He stared at the framework, because he still had no idea what to put inside it.

I told him to try a different question: “What does an interesting story look like?”

This time ChatGPT showed him several examples – vivid scenes, concrete details, emotional turns. It clicked instantly – oh, so that’s what “interesting” looks like. Looking back at his own “the play date was fun” essay, the gap was obvious.

Same tool. Asking How produced a useless methodology. Asking What produced a standard he could benchmark against. The difference wasn’t the tool. It was the person asking.

Understand the Problem Before Solving It

My son’s problem wasn’t that he couldn’t write. He had no idea what an “interesting story” looked like. Without a standard, how could he possibly produce one?

This made me realize a very common thinking habit: when facing a problem, the instinct is to ask “how do I do it” rather than first clarifying “what does done well look like.”

How gives you a sense of action – once you start “doing,” the anxiety eases. But the quality ceiling of any How isn’t determined by the How itself. It’s determined by how clearly you understand the What – the definition, the standard, the picture of “done well.”

What caps How – the quality ceiling of any output is set by the precision of your understanding of What.

Want to lose weight? What does “thin” mean? A certain number on the scale? A body fat percentage? Fitting into certain clothes? If your What is just “I want to be thinner,” you’ll bounce between diets endlessly, because without a standard, you can’t judge which How is right.

Choosing a school for your kid? What’s a “good school”? High admission rates? Close to home? Teaching philosophy aligned with yours? Everyone’s definition differs, but you need your own definition first, or visiting ten schools will only leave you more confused.

Want to write a great article, build a great proposal, deliver a great product – what does “great” actually look like? Without a clear standard, all the techniques and tools in the world are a gamble.

You Think You’ve Figured It Out – You Haven’t

The hard part isn’t knowing you should clarify What first – most people know that. The hard part is that What has levels of precision, and people too easily deceive themselves at a low-precision What.

Level one: labels. “Really fun.” “I want to lose weight.” “I want to build a good product.” You’ve slapped a category on it – almost zero useful information.

Level two: descriptions. “Playing games with friends was fun.” “Lose 10 pounds before summer.” “Build a product with high user retention.” There’s a direction now, but it’s still vague.

Level three: scenes. “The ten-second screaming moment when we pulled off a comeback in the final round.” “Drop body fat from 25% to 18% and fit back into last year’s pants.” “New users complete the core action within 30 seconds of first open; 7-day retention hits 40%.” At this level, the How practically surfaces on its own.

Most people start executing at level one. A few get to level two. People who push What to the scene level look like they have strong execution and decisive action – but they don’t. It’s that once What is clear, How becomes obvious.

Sharpen Your What with Why

How do you push What from a label to a scene? Ask Why.

Why isn’t a “step two” after What. It’s a whetstone – keep asking Why until your What is sharp enough.

“I want to lose weight.” – Why?
“Because I feel fat.” – Why do you feel fat?
“My pants from last year don’t button up.” – So your standard isn’t a number on a scale. It’s fitting back into those pants.

Three Whys later, What has gone from “lose weight” to “fit back into those pants.” The latter is specific enough that you can try them on weekly to track progress, while “lose weight” just leaves you anxious in front of a scale.

This is the same logic as Toyota’s 5 Whys – dig a few layers below the surface problem to reach the real one. You think you know what you want, but after a few Whys you often discover that what you actually want is nothing like what you originally said.

In the AI Era, What Is the Only Moat

Back to my son’s two queries. Same AI – asking “How do I write an interesting story” produced an empty framework; asking “What does an interesting story look like” produced a benchmarkable standard.

AI is a How-amplifier, but it can also be a What-clarifier – the prerequisite is that you know to ask What.

The bigger picture: AI is driving the cost of acquiring How toward zero. Writing, proposals, analysis – Hows that used to require years of training can now deliver a decent result with a single prompt.

When everyone can get an equally good How, the only differentiator left is What.

Whoever can define the problem more precisely, whoever can describe “done well” more clearly, gets better output from AI. This isn’t a technical skill. It’s a thinking habit.

If someone habitually asks AI “how do I do this” for everything, they’re training their ability to invoke – while atrophying their ability to define problems. Over time, they turn themselves into an AI wrapper – input in, output out, no judgment of their own.

A Self-Check Habit

When you catch yourself asking “How do I…,” pause. Ask yourself:

“What does done well look like? Can I describe a specific scene?”

If you can’t, keep asking Why – why am I doing this? Why now? Why does it matter? After a few rounds, What will clarify itself.

Once What is clear, How surfaces naturally. In the AI era, you don’t even need to come up with the How yourself – but What can only ever be defined by you.

Back to my son. The essay he turned in that day was leagues better than the first draft – not because ChatGPT taught him any writing techniques, but because he finally knew what “interesting” looked like.

The tool didn’t change. The question changed. And the output changed with it.

What caps How.