$ cat ai-made-building-fast-thinking-is-scarce.md

AI made building fast. Thinking is the scarce skill now.

· #ai #developer-productivity #system-design

Graphic titled "How to stand out in an AI world by thinking better, not faster," showing a person looking at a glowing brain surrounded by questions like "What is the real problem?" and icons for problem framing, architecture, understanding, and judgment, over a quote: "In an AI world, the person who wins isn't the one who ships faster. It's the one who creates value."

AI is making us faster. The question I don’t hear asked enough is whether it is also making us think less.

(This started as a LinkedIn post.)

More and more, I see people take LLM output at face value: ask, generate, ship, repeat. That loop is incredibly fast. But speed without judgment produces a different problem, more output and weaker thinking.

What the time is actually for

My own mindset changed because of AI. It now writes a lot of the code I used to spend hours on. The real advantage isn’t that I write more code. It’s that I get more time for the parts a model can’t do for me: understanding the actual problem, framing it correctly, designing the architecture, reasoning about how the system behaves, and making decisions that still hold up a year later.

Building fast is becoming easy. Building something useful, maintainable, and still coherent in a year is not. A product can ship in days and become a mess to maintain. A feature can look complete and solve the wrong problem. A technically correct AI-generated answer can still be the wrong decision.

That’s the same reason I read every line of the code AI writes for me. The speed isn’t there so I can merge more; it’s there so I can spend the saved time on the parts that need a human.

The skills that get more valuable, not less

The better AI gets at execution, the more the scarce work moves upstream of the keyboard:

None of these are what a model does when you ask it to generate something. They are what decides whether the thing it generates was worth generating. AI should help us move faster after we understand the problem. It shouldn’t replace the thinking required to understand it.

So the person who wins in an AI world isn’t the one who ships faster. Building more is about to be achievable for almost everyone. Building the right thing, and making it last, still requires us to think.

← cd ../writing