Writing with an LLM

Summer 2024, AI chatbots were becoming a very big deal. Everybody was talking about using them to generate code, text and images. It gave me the motivation to start writing a tech blog again, after maintaining one for years at the beginning of my career.

I always liked taking notes on my ideas, arranging them in my mind, then putting them into words. But I never really enjoyed the tedious editing part where I try to make these words appealing to read. It’s the most time-consuming part, and I can’t offload it to my subconscious while doing something else.

So I gave ChatGPT a try to write some of the first articles here.

But I always kept one base rule: all ideas come directly from my brain, as well as the structure of the article. I dump all the ideas into a draft, then ChatGPT rearranges them to make something that looks professional.

It really helps when you are not a native speaker of the language (mine is French).

The result is satisfying at first. My ideas are here, and it reads well.

But it’s also very different from my writing style, which is minimalist and concise.

And these days in 2026, AI-generated content (aka slop) is increasingly hated, and people value anything purely created by a human. I feel the same.

Now my rule is clear → I only allow an LLM to review my articles, and decide which suggestions to apply.

I make the changes myself in the text. It’s the good old way to memorize and internalize the fixes for my mistakes: grammar, style, wrong ideas. It can also help me clarify my thoughts.