@beckermatic Maybe the person you are blaming is overworked and/or has been improperly trained, maybe they are severely sleep deprived, and shouldn't even drive, but have to. You'll want to look up for causes before you blame and punish someone. I'm sure this worker is devastated by their mistake.
Hats off to the author, you don't see that kind of, uh, skillful rhetoric chicanery every day. Like "generative AI doesn't compete with artists because artists are not in the data market". 😬
@scy@creativecommons I think it's saying that training the model is fine, not plagiarism, using the model is quite different though, as it will absolutely produce things competing in the aesthetic market of the originals.
So, over the past few weeks, there has been a lot of interest in the protests around the country as a form of "social contagion." First of all, some of this pathologizes what is happening as akin to an infectious disease. 1/
@Melissabeartrix i'm sorry but they don't, sometimes they get along, sometimes they don't, we shouldn't idealize nature because we managed to isolate from it, it's brutal out there. Still, we are in position to do better.
On a related note, and i'm sorry to bring it up, but i believe you aren't vegan.
@Melissabeartrix Oh well, that's nice to know, it would be nice if more people aimed for that, going 100% is a lot harder than even 99%, and i'd rather fail sometimes but still keep at it, than give up because it's just too hard. Going most of the way does a massive difference of impact. So cheers for that 👍 .
Le procès de l'ancien président américain en Floride, accusé de détention illégale de documents classifiés après avoir quitté ses fonctions, a été reporté. Le juge Cannon a suspendu le processus du 20 mai sans fixer de nouvelle date.
Judge indefinitely postpones Donald Trump's classified documents trial, saying issues in presenting classified evidence in court need to be resolved.
ABC News reports it's now unlikely the trial will start before Election Day. "Trump pleaded not guilty last June to a 37-count indictment related to his handling of classified materials. Prosecutors said he repeatedly refused to return hundreds of documents containing classified information ranging from U.S. nuclear secrets to the nation's defense capabilities, and took steps to thwart the government's efforts to get the documents back."
@Nonilex@kwheaton sadly they quickly forget what they need to, it’s not the first or even the 1000th terrible thing that should have killed him politically.
@gknauss never had any issue giving, but many people gave up after fainting, or never even trie in the first place, so if you hurt yourself before, it's indeed an accomplishment to go over that and give again, this time with a better experience.
It's been some time i didn't give, i need to go. So thanks for the reminder. 😄
I started a project for searching Unicode characters from your terminal, similar to unicode.party and @sethmlarson's utf8.xyz (but as a TUI instead of a web app).
I initially thought I'd call the project "utf", but that has a name conflict on PyPI (due to similarity to utf7, utf9, etc. likely).
La communication de crise de Tefal, c'est pas ouf. Je veux dire : qui se sent rassuré par un slogan aussi plat que "Nos revêtements sont reconnus comme sûrs" ?
On dirait le vieux gag de Boulet sur les publicitaires qui doivent vendre du surimi, et qui, pour ne pas faire de la publicité mensongère finissent par proposer le slogan "une intéressante source de protéines"...
Ce qui est intéressant, dans cette affiche, c'est ce qu'elle ne dit pas, justement...
@Emmaf_77
C'est vraiment compliqué d'affirmer plus non? On peut dire qu'il n'y a aucun problème connus (donne le cadre d'une utilisation normale, oui, pas si on mange la poêle j'imagine), mais que c'est sûr ? C'est démontrer l'inexistence de risques. C'est logique qu'ils restent aussi vagues, mais c'est clair qu'on aimerait une affirmation plus qualifiée @LegalizeBrain
@mamund a classic move, regulation will create a barrier entry for competition, while they'll have an easier time to adapt to, or even shape, the regulation they to benefit them.
@grebert j'imagine que les gens qui achètent neuf peuvent se le permettre, et ces dernières années la taille et le poids des voitures ont fait un bond, et le prix avec, je pense que les gens qui gagnent 100k+ s'en préoccupent peu. Par contre, ça va faire mal sur le marché de l'occasion quand il n'y aura pratiquement plus que ces véhicules hors de prix.
Mais bien content de ne pas avoir besoin de voiture, et pas seulement pour le coût.
i need to keep a count of the times where i see people studying chatgpt mistakes asking chatgpt to explain its reasoning or to reflect on the mistakes
IT CAN'T
IT HAS NO INTERNAL WORLD MODEL OR CONSCIOUSNESS OR REASONING PROCESS
so it's not explaining its mistakes to you, its generating something that looks like an explanation but there is no reason to believe that it's connected to the actual process that generated the mistake
@fay i agree about the consciousness or reasonning process, but model of the world? i don't know, i've seen experiments that indicate that some models do represent a chess board internally when you play chess with them (it wasn't chatgpt, but it was a variant of gpt), so although it doesn't mean it generalize to the things people try to "discuss" with it, i wouldn't affirm that it has no model of the world. Complex enough rules about text tokens might end up being rules about the world.
@fay i'm indeed not sure what the categorical distinction would be. Some models are better for sure, their rules encode more fundamental aspects of the thing they describe, rather than fuzzy inferences, but i'm not convinced that with enough training, the kind of machines we build, can't arrive at such a model. Not saying it's practical.
But of course, that doesn't necessarily make these models good or useful, or the best solution to do the things we want to do with them, even when they work.
This map shows where Russia is currently attacking GPS signals in the EU.
According to this brilliantly researched article from Peter Møller, Russia has a classified weapons system called Tobol, which is jamming GPS satellites from a base in Kaliningrad.
By using photos and articles from before the war, he has managed to piece together a surprisingly clear picture of what the Russians are doing.
(In Danish but translating it is absolutely worth it.)
@elpolacodesplegado@collectifission@randahl well, if EW is considered to be hostile, should we consider ourselves under attack? If so, maybe we should at least make that known to the kremlin.