@eniko maybe an unpopular opinion based on the replies here, but the generalization of LLM's and "AI" in this thread makes these absolutist statements against "it" hold way less ground.
You specifically mentioned "cloud based" at the beginning, leaving the potential for supporting locally hosted LLMs, but then started using the term AI more generally. So that also makes it a bit harder to nail down what you're actually against here.
If the idea is that LLM's and, in general, all machine learning algorithms are inherently bad by the fact of their mere existence regardless of how they are created, executed, or used, then I think you've actually stumbled into a bit of a hard line stance. If I'm wrong about this, and it wouldn't be the first time I've completely mistaken somebody's opinion, then please let me know.
If I'm on the right page, then I'm curious what you consider the line the seperates okay-to-use Algorithms versus ones that will always be bad. Is it the type of algorithm, the way it's used, the type of data it's using for input, the type of data it outputs, does the context matter, am I completely missing something here?
Hope that wasn't too much for a single reply. Still not used to platforms like this.