I genuinely believe if the author understands it, AI tools to assist with development are amazing. I use an in IDE AI autocomplete, and it has increased my productivity dramatically without it. I'm still very capable without AI autocomplete, once you start to over rely on AI tools, and lose understanding of your codebase is when you're more vibe than not, and that's where I personally take issue. AI generated code I think is mostly fine assuming it isnt just a copy and paste from a oneshot chatgpt message with absolutely no understanding of what the code is doing whatsoever.
The interesting thing here is not their policy -- it won't survive, since they've given themselves the task of judging random people's skills -- but the fact that they have an AI policy for their open source project.
I think every project will need one, not least because of the issues/PRs full of AI slop.