AI-Assisted Building

What I stopped Googling, and what I still ask people

AI replaced my search bar for a whole class of questions. It hasn’t replaced the ones only a human who knows the context can answer.

What I stopped Googling, and what I still ask people

The search bar I stopped opening

Somewhere in the last year, I noticed I had mostly stopped Googling a certain kind of question. How do I write this bit of syntax. What does this error actually mean. How does this API want its arguments. Questions with a single definite answer, I now ask the machine, and it’s faster and kinder than sifting through ten forum tabs and a decade-old accepted answer that no longer applies. But I noticed something else too. There’s a whole set of questions I still bring to people, and I always will.

Two kinds of questions

The line turned out to be clear once I saw it. AI is excellent for questions with a knowable, general answer, the things that are written down somewhere in the world and just need retrieving and shaping. It’s no substitute for the questions that depend on context, relationships, and lived judgment, the ones where the answer lives in a specific human’s head and nowhere else.

  • Ask the machine: syntax, definitions, how a known thing works, a first draft, an unfamiliar concept explained.
  • Ask a person: is this right for this client, what’s really going on in this team, what would you actually do here.
  • The machine knows the general case. People know your case.
Use AI for the answers that are written down somewhere. Use people for the ones that only exist in context.

Why the distinction matters more, not less

As the lookup questions get answered instantly, the judgment questions become the whole value. When anyone can retrieve the general answer in seconds, what you’re actually paid for is the contextual one: the read on this specific situation that no model has seen. That’s the same reason AI is a great junior and a dangerous senior, brilliant at the retrievable, out of its depth on the situated. And it’s why the discipline of asking well, of treating the prompt as a spec, matters: a machine can only answer the written-down part, so knowing which part is written down and which part is yours becomes a real skill.

I use AI constantly, and I wrote up exactly where I draw that line day to day in how I actually build with AI. Stopping the lookups didn’t make me need people less. It sharpened what I go to them for.

What this means for how you learn

Let AI take over the lookup questions, gladly, it genuinely frees hours. But keep bringing the judgment questions to people who know your situation, because that’s exactly what the machine can’t know, now or later. Knowing which question goes where is quietly becoming one of the more useful skills there is. If figuring out where AI fits in your own work is on your mind, I’m always glad to compare notes.

Related reading

Keep reading

Have a project in mind?

Let's build something that lasts

Get in touch

Based in Manila, working with teams across time zones.