Write the damn code

Here's some popular programming advice these days:

Learn to decompose problems into smaller chunks, be specific about what you want, pick the right AI model for the task, and iterate on your prompts.

Don't do this.

I mean, "learn to decompose the problem" — sure. "Iterate on your prompts" — not so much. Write the actual code instead:

• Ask AI for an initial version and then refactor it to match your expectations.
• Write the initial version yourself and ask AI to review and improve it.
• Write the critical parts and ask AI to do the rest.
• Write an outline of the code and ask AI to fill the missing parts.

You probably see the pattern now. Get involved with the code, don't leave it all to AI.

If, given the prompt, AI does the job perfectly on first or second iteration — fine. Otherwise, stop refining the prompt. Go write some code, then get back to the AI. You'll get much better results.

Don't get me wrong: this is not anti-AI advice. Use it, by all means. Use it a lot if you want to. But don't fall into the trap of endless back-and-forth prompt refinement, trying to get the perfect result from AI by "programming in English". It's an imprecise, slow and terribly painful way to get things done.

Get your hands dirty. Write the code. It's what you are good at.

You are a software engineer. Don't become a prompt refiner.

★ Subscribe to keep up with new posts.