Why would they add LLMs to Calibre?

It looks like it's so you can "discuss" a book with the overgrown autocomplete, or ask it what to read next if you are incapable of independent thought. I could, sort of, understand making a plugin for this sort of nonsense, but building it in, as a brief check of their site suggests has happened, is so utterly ridiculous that it defies rationality.

I liked Calibre, I've even contributed code to it, but it looks like I'll be investigating this fork now.

Why would they add LLMs to Calibre?

So it can scan text.

I have not heard of this program until just now, but presumably they're not integrating an LLM so you can stop and chat with fictional characters. It's gotta be an overpowered answer to 'here is some plaintext, populate its metadata.'