Comment by zmmmmm

They compare it to software development but there is such a crucial difference to software development: by and large, software is an order of magnitude easier to verify than it is to create. By comparison, reviewing a vibe generated manuscript will be MUCH more work to verify than a piece of software with equivalent complexity. On top of that, review of academic literature is largely outsourced to the academic community for free. There is no model to support it that scales to an increased volume of output.

I would not like to be a publisher right now facing the enslaught of thousands and thousands of slop generated articles, trying to find reviewers for them all.