I believe they're all in different "altitude" orbits, so you'd need a pretty large amount of delta-V (read: fuel) to get from one to another, far more than the characters would have had available.
It sounds like you continue to use it, though. How do you justify it in the face of what I laid out above? "Waste of energy" is a shitty excuse for engaging in bad behavior.
There was a post on Mastodon that I sadly cannot find right now that really articulated the fact that there's not necessarily a single problem with LLMs and generative AI - the issue is that there's an entire stack of potential dangers associated with them. To paraphrase:
Use of and reliance on LLMs for certain tasks has shown to have deleterious effects on critical thinking skills.
Even if that isn't true or I weren't concerned about it, I'd be concerned about its effects on my psychological wellbeing.
Even if I weren't concerned about that, I'd be concerned about the ethical issues of how their training data was and is acquired.
Even if I weren't concerned about that, I'd be concerned about its effects on the job market and the further upward concentration of wealth.
Even if I weren't concerned about that, I'd be concerned about the massive energy costs and the associated effects on utility bills and greenhouse gas emissions.
Even if I weren't concerned about that, I'd be concerned about the massive cooling requirements and its effects on the global availability of clean water.
Even if certain approaches to or implementations of GAI solve one or a couple of these concerns, I'd have to overcome all of them (and likely others I've forgotten to list) to feel comfortable using GAI in any serious capacity, and even then it looks like I would end up with a tool that I'd have to constantly double-check to avoid hallucinations. It's just not worth it.
And nearly all of these arguments also to apply to others using GAI, so I'm forced to advocate against it.
Edit: the vulnerabilities mentioned below were all implementation-dependent; the protocol appears to be fine.
I haven't been following Matrix development too closely, but last I heard, both the protocol and the reference implementation had serious flaws, including gaping security holes. As in, issues that couldn't be overcome without a clean-slate redesign. Did they somehow manage to salvage something useable?
A fur of taste.