Outsource the Syntax, Keep the Logic: How I Use AI Without Losing My Thinking

Outsourcing our thinking to AI is going to create an interesting future.

Calculators made maths easy, and yet many of us now struggle to do it in our heads. I suspect thinking will go the same way: the less we use it, the less exercise our brains get, and the harder thinking becomes.

So far I’ve tried not to use AI for my thinking, but as another mind alongside it. I enjoy finding solutions to problems. Outsourcing that would feel like losing a whole part of me.

Two kinds of work in every problem

Since I code a lot, I separate AI use into two kinds: syntax and logic.

Syntax is a language

There’s a vocabulary, and there are rules for building a sentence. Every programming language has its own. The same idea, looping over a list, looks different in each:

# Python
for item in items:

// C#
foreach (var item in items)

// JavaScript
for (const item of items)

For syntax, AI is genuinely powerful. You don’t have to learn an entire language, or know all its ins and outs, to do something useful with it. Often you can read it from its similarity to languages you already know. AI is a great translator.

Logic is communication

You can have all the words and still not know how to use them. That’s why some people are great salespeople: we all speak English, but they know how to use it to sell. Thinking works the same way. Framing a problem, breaking it into parts, knowing where to focus, working out how to solve it: these are skills you build, and they matter.

Why outsourcing logic is risky

If we outsource logic, we become poor problem solvers. If the servers go down, life ends. And depending completely on anything is never good. Relying on AI to always be right is risky too: it can make things up, or act like a yes-man that simply reinforces your idea as if you’re always right.

Some thinking is worth outsourcing

That said, not all thinking is worth the effort. Picking a shirt every morning feels like a chore to me, so I wouldn’t mind outsourcing it. For a fashion designer, though, that’s exactly the thinking not to outsource, because it’s their creativity. The rule I’ve landed on: outsource what isn’t your craft, and protect what is.

How I actually use it

The balance that has worked for me is to use AI to find the holes in my thinking:

1. Paint the picture and explain your approach.
2. Ask it to be an ENTJ, a blunt and challenging personality type, and hit you hard with real questions and counterarguments.
3. Do the analysis and problem solving yourself, and use the back-and-forth as checks and balances until you reach something that works.

Want more? Take the whole thing to a second AI. I did that once for fun, and it was funny watching them keep telling each other they were wrong.

How do you use AI in your own work?