[fbnlsr]

This has been my experience as well.

- Claude writes User Stories, supervised by the PO.

- Claude is in charge of the implementation, supervised by the devs.

- Claude does the PR review.

- If a comment is made by a human, someone c/p what Claude thinks with a simple "not sure if AI is right".

We're just passing butter at this point.

[glaslong]

My experience as well. It's happened a few times now that I'll ask a coworker a question about some system they built, and they tell me to just ask the AI.

I did ask the AI. I gave it several rounds. I got a summary of their diff from it. I skimmed their diff myself. I have a rough mental model of the work they did, plus what the AI did/didn't tell me. I'm asking THEM so THEY can confirm my mental model is correct to the code they supposedly authored.

I suspect the real problem is they vibed the whole thing and didn't even self-review it. So they don't know, can't answer, and the interaction is truly pointless. They are not the expert any longer. It's not worth asking them.

[indoordin0saur]

Currently having a big issue in some of our data pipelines because a lazy engineer vibe coded a bunch of slop that doesn't make much sense and the AI can't help. Looking like we're going to need to scrap all the code and start from scratch.

[Jianghong94]

You know the funny thing is that, the lazy engineer can very likely ask you to just scrap all his code and vibe code again.