Agree, but if we are going towards the AI writes all code future that these AI companies want, then code review will become inneffective.

If we stop writing code, we will lose that "edge" that helps us detect bad logic when reading a PR.

The more code we have to review, the more mistakes we will make letting bad code through (I believe this is the case). It's less fatal when the code we review is written by smart humans because it's unlikely they code something as bad as an AI can produce.

Anyways, I think using any AI agent to write bigish blobs of code is a mistake, but if you are gonna do it i hope you have multiple fresh pairs of eyes on each PR. Still will give me the ick seeing an AI PR get merged tho.