The difference probably is that GCC extensions have been stable for decades. Meanwhile Rust experimental features have breaking changes between versions. So a Rust version 6 months from now likely won't be able to compile the kernel we have today, but a GCC version in a decade will still work.
How does OpenAI being liable for reproducing copyrighted material imply that a word processor should be as well? Last time I checked, word processors don't have a black box text generator trained on pre-existing works: a word processor only has the text that the user types into it.
> Not really. Youtube is not liable as long as they remove the content after a copyright complain and other mechanisms.
They have to take action precisely because they're liable for the material on their platform.
> I don’t think a country’s government can justify no commercial LLMs to its populace
They're not saying no LLMs, they're saying no LLMs using lyrics without a license. OpenAI simply need to pay for a license, or train an LLM without using lyrics.
But lyrics are just one example. Are you saying that training experiments must filter out all substrings from the training input that bear too close a resemblance to a substring of a copyrighted work?
Obviously there's a limit, reproducing a single sentence is unlikely to be copyright infringement just because there are only so many words in a language; but if reproducing some text would be copyright infringement if a human did it, I don't see why LLM companies should get a free pass.
If it's really essential that they train their models on song lyrics, or books, or movie scripts, or articles, or whatever, they should pay license fees.
That is a separate opinion, but with respect to the question at hand, the utilitarian value of being able to ask a computer "what are the lyrics to x" and having it produce them outweighs whatever small ideological sanctity the music labels assign to being able to gatekeep the written words of a composition to a small blessed few. It's not like chat gpt is serving up the mp3 file to you. So correct, it is insane to me that mere reproduction of just the lyrics is afforded such weighty copy protection.
(Vis a vis, I take it you write a certified letter to Universal before reproducing Happy Birthday in public? ;) That is actually a far more egregious violation indeed, as it is both a performance of the copyrighted work and in front of an audience - neither of which are the case for the chatbot - yet one we all seem to understand to be fair use.
This obviously applies to all copyrighted works. I could sue OpenAI when it reproduces my source code that I published on the Internet.
They already "filter" the code to prevent it from happening (reproducing exact works). My guess it is just superficially changing things around so it is harder to prove copyright violations.
Buy your steak, toss in a pot of unsalted water. Cook for a while to make it edible. Eat it when hungry. Tell me again how tasty the meat is.
Do the same with rice, potatoes or lentils and you'll have completely different experience. Pick any fruit. There's even no need to boil. Tasty from the get go.
That's a good point, I've personally not got much use out of LLMs (I use them to generate fantasy names for my D&D campaign, but find they fall down for anything complex) - but I've also never got much use out of StackOverflow either.
I don't think I'm working on anything particularly niche, but nor is it cookie-cutter generic either, and that could be enough to drastically reduce their utility.
> I want to be clear that I am generally limiting my scope here to rigid non-powered armor. Power (or powered) armor – that is, armor that moves with built-in servos and motors, rather than purely under muscle power – is its own topic that we’ll leave for another day.
Ah, fair, I think. I confess I assume that there is some sort of power source on all of the ones discussed, though? The Dune setups clearly have a power source for the shield stuff. Why wouldn't that power source also work to help move the armor?
reply