The problem is, LLMs (Ai) isn't just another tool. Even if the code is reviewed well and not slop at all, there are inherent uncertainties using such tools to generate code. We don't know how the future of code generated by Ai can be licensed and used; it does not even provide the original license of the code it was trained on. This is not just ideological, there are unsolved risks. And this is the best case scenario, as we assume the developers aren't slopping around.
Mind you, I'm critical about Ai, especially the online tools. But I'm not entirely against it, yet. I can however understand why people are against using Ai in Linux is a critical issue, for the reasons I described above. For myself I just play around with local Ai (on my 8gb VRAM+32gb RAM btw), so you see I'm not entirely against it.