You are missing the point

> AI stream is a lot higher than what human review can take seriously, leading to either a longer list of un-reviewed PR's or not fully reviewed code and quality issues. Both thing we don't want in the project.

It’s not about detecting AI, it’s about reducing the amount of code to review.

▲

XorNot 2 hours ago | parent [-]

Okay for feature work but the amount of bugs I can find with Claude in hour, with deterministic reproducers is staggering.

	▲	dspillett 22 minutes ago \| parent \| next [-]
		If you are reporting a number of bugs per hour then you can't be properly reviewing each to make sure it isn't a complete hallucination, a behaviour-as-designed reaction to silly configuration, etc, nor are you reviewing the documentation to make sure it both actually makes sense and is appropriately concise to not waste time in the review process. This sort of thing, in the minds of not just you but many other people, is exactly the problem. On behalf of the projects that you have communicated with this way: thanks a bundle.
	▲	croes an hour ago \| parent \| prev [-]
		Bug reports also need review because not every bug reported is a real bug, could be hallucination and the "fix" is useless or even dangerous. You don’t know the model and prompt they used to generate the report so you can’t even take the model quality into account