On the one hand, Toybox exists. So, the non-copyleft license bs isn't new. On the other hand, toybox afaik isnt aiming to treat "deviations with GNu as bugs". Almost feels hostile-takeover-ish though I know that almost certinly isn't the idea behindbit.
If this ends in proprietization bs I'm going to throw hands.
That's a lot better than it could be. But I'm also talking about training costs. Models have to be updated to work swimmingly with new languages, conventions, libraries, etc. Models are not future-proof.
There are more efficient training methods being employed. See: the stuff R1 used. And existing models cam be retooled. But it's still an intrinsic problem.
Perhaps most importantly it's out of the reach of common consumer grade hardware to train a half decent LLM from scratch. It's a tech that exists mostly in the scope of concentrated power among peoole who care little for their enviromental ramifications. Relying on this in the short term puts influence and power in the hands of people willing to burn our planet. Quite the hard sell, as you might imagine.
Also see: the other points I made
I aint just talking about inference. Training costs are insane and models have to be updated to be used well with new languages, libraries, etc.
Energy and water costs for developmenr and usage alone are completely incompatible with that. Come back in 20 years when it's not batshit insane ecologically.
Not to mention reducing power usage of programs isnt going to be very feasible based on simply an LLM's output. LLMs are biased twoards common coding patterns and those are demonstrably inefficient (if the scourge of web apps based on electron is any tell). Thusly your code wouldn't work well with lower grade hardware. Hard sell.
Theoritically they could be an efficient method of helping build software in the future. As it is now that's a pipe dream.
More importantly, why is the crux of your focus on not understanding the code you're making. It's intrinsically contrived from the perspective of a solarpunk future where applications are designed to help people efficiently - without much power, heat, etc.. weird man
True. I wish expectations aligned more with this kind of practice... I see a lot of people thinking in binaries about these kinds of issues: that either we should just trust (insert company here) completely or distrust... People should learn to demand accountabillity and transparency instead of just eating corporate drivel or becoming skeptic of Anything Ever.
What in the fuck?
It just kept going. Calling customers "zombies"..... bannung people from using chrome to access website, banning "regular reddit users"...... banning particular countries for pettt grudges with like, a few individuals???
Not the OP but I find it really curious you trust Apple's custom silicon over Google's in such a way that you seem rather confident with your locked down, Apple-controlled OS... unless you root it but then GL getting security updates, which kinda defeats the whole point.
There are a lot of risk vectors and frankly if we're getting to the immutable silicon level the exploits are probably going to be filed under direct targeting level privacy concerns. GrapheneOS is really mature and has security experts so good at their jobs they occasionally apply security patches before Google, with all its funding, does.
Apple and Google were both caught providing push notification data (e.g. the service underpinning most of the notifications appearing on your device) to the authorities. Not to mention Apple argued in a court case that it's "unreasonable" for an Apple user to presume their activities are private from Apple. I believe that particular court case was about them constructing ad profiles on their users.
It's genuinely absurd to me that you just vaguely threw out some kind of blanket concern with Google's underlying hardware. I'd really love to hear if that's based on anything material. From information I know is real and privacy concerns that have been explicitly justified, GrapheneOS is the best choice on the market atm.
Sunshine (the server) and moonlight (the client)
Eh. Fediverse is a portmanteau of federation and universe. It's perfectly fine and not misleading imho to suggest other federation-capable services in this context. Especially considering Matrix is definitely part of the wider cultural context underpinning the transition of some groups to using these counterculture services (it's often frequently recommended in these contects, often used by people on mastoson etc anyways),
At this point we all need fake social media accounts