The Loss of Code Ownership

I am noticing a trend where I and other developers rely on AI more and more and more and the problem here is that work gets completed but we don’t do a good job of actually checking that the work is actually complete, that it fully implemented the requirements. This is fine in a personal project (who suffers but myself?), but its a problem in corporate because the person who suffers is the poor sod QAing the work.

As a developer, I accept that work produced by my agent is MY work and therefore I am at fault if it is incomplete, poorly implemented, or buggy.

But there’s a bit of a mental disconnect because it doesn’t feel like MY CODE anymore. I was just lamenting this online. I no longer know my code deeply. It used to be that I’d strive for the perfect elegant solution and the resulting product would be MINE, and I’d know the flows inside out. Now I have to talk over issues with (agent) Bob to find where the problem may be and how to implement new features. I review plans, I approve them, they are implemented, I review them, but I don’t know them in my soul the way I would if I had written them.

Most days this is fine and I don’t think about it. Today it feels a bit joyless. I’ve avoided leadership positions of a higher level where I no longer interact with the tech because I like getting my hands dirty, but I’m not really getting my hands dirty anymore. I’m a chief architect that sets the rules and the architecture and monitors it from a distance.

And yet I wouldn’t want to go back. It was too slow by comparison. Hmm.