Serguey Asael Shinder – IT Professional

When the tools got good enough to write real code, I noticed a subtle temptation in myself. If the model wrote it, some part of me felt a little less responsible for it. If it turned out wrong, well — the AI generated that. It was a comfortable, and completely false, way to think.

Because the commit has my name on it. Not the model’s. And “the AI wrote that part” is not a sentence anyone gets to say at an incident review with a straight face. The responsibility never transferred when the typing did. It stayed exactly where it always was — with me.

The model has no skin in my game. It won’t be paged at three in the morning when the thing it generated falls over. It won’t sit in the retro explaining what went wrong. It won’t carry the weight of a mistake into the next project. I will. All of that is still mine, no matter who or what produced the first draft.

So I made myself a rule that keeps me honest: read everything the model gives me as if I’ll have to defend every line — because I will. Not skim it, not trust it because it looks fluent. Defend it. If I’m not willing to stand behind it, I’m not allowed to ship it.

I use these tools every day and I’m faster for them. But the ownership didn’t move. As far as anything that matters is concerned, I wrote every character. Working that way is the only version of using AI I can actually respect.

– Serguey Asael Shinder