You can run a lot of AI stuff on your own hardware offline, check the catalog for /lmg/ and /ldg/ for a starting point.

I'm glad I jumped off of fedora this year, I'm just sticking to debian now

>>108008299
even the local apps are invasive with telemetry or saas grifting. we need a projects that don't touch the web at all

>>108008299
as opposed to?
you want gemini to to fold your laundry and gpt to do your dishes? robots will be too expensive for most, you're better off hiring a mexican nana to do those for you

Why do people make OPs with images that are only tangentially related to the topic?

>>108008355
As opposed to running locally, where you don't have to constantly buy more tokens.

>>108008337
https://rentry.org/IsolatedLinuxWebService or running the program with unshare -r -n will castrate all the telemetry, if you don't want to see saas shilling don't use comfyui and use forge neo or sd.cpp instead, llama.cpp or kobold.cpp for text.

>>108008564
you have crap ton of local models what are you talking about

>>108008299
>online-only
The same reason most games today require online connectivity, or microslop forcing you to use an onlne account:
It's easier to monetize.

>>108008299
llms require 1tb of ram

>>108008299
Bye bye quality!

>>108008321
>I'm glad I jumped off of fedora this year, I'm just sticking to debian now
Good, so you just wait 2 years, for the AI code to be merged on debian. 90% of programmers already use AI assistance, the good thing about Linux and FOSS in general is: people contribute out of passion, only a few elite programmers are paid and they don't need AI. Chill anon.