Yeah i remember that Ed article ! I don’t think the technical aspects are relevant to the newer generation of models, but yeah of course any attempt to compress inference costs can have side effects : either response quality will degrade for using dumber models, or you’ll have re-inference costs when the dumb model shits its pants. In fact the re-inference can become super costly as dumber models tend to get lost in reasoning loops more easily.
This comparison is really common but i totally disagree with it.
NFTs are a total invention, the poster child of a solution looking for a problem. Nobody, anywhere on earth, ever wondered how they could get their hand on procedural art and speculate on it. The whole metaverse thing was the same way. The pure product of people who are so detached from normalcy that they can’t even begin to fathom what humans tend to like and dislike in the real world.
AI has a million problems which don’t need reiterating. I’m not disputing any of that. It may or may not be a technology that’s viable in the long term (economically and environmentally), i won’t speculate on that. But pre-LLMs there was a huge demand for better natural language processing. For semantically aware programs that are able to generalize and don’t need retraining every 4 days. It was kind of the final frontier for software, the limit between “i can do that in a few sprints” and “i’ll need a bunch of PhDs and 2 years of runway to possibly maybe make something work”.
And i understand that final users only see the dogshit copilot integrations that they never asked for, and which are being pushed to their devices against their will, and becoming a privacy nightmare. And i understand that it’s tiring to hear brain-rotted maximalists constantly making up idiotic predictions about humanity’s future while they let the “groundbreaking tool of the day” riffle through their inbox and bank statements. But it would be a categorical error to believe that LLMs are anywhere as useless as, say, NFTs or the Metaverse.
Yeah that’s also something that you have to train for, i’m not super aware of the technicals but model routing is definitely important to the AI companies. I suspect that’s part of why they can pretend that “inference is profitable” as they are already trying to squeeze it down as much as possible.
I hate the “Random industry is cooked” trope. Right now if you go on LinkedIn you’ll find a ton of bros proclaiming the death of Hollywood because Seedance can now generate 3 minutes of incoherent eye-candy with vaguely realistic looking special effects.
It’s like brainrot for boomers, you can see they are getting absolutely hypnotized by the loud noises and bright colors, and totally missing that the main character who was running to the right with blue pants is now walking to the left with red shorts.
I mean what’s the point of morging if you’re gonna do it intermittoucly?
To clarify : model collapse is a hypothetical phenomenon that has only been observed in toy models under extreme circumstances. This is not related in any way to what is happening at OpenAI.
OpenAI made a bunch of choices in their product design which basically boil down to “what if we used a cheaper, dumber model to reply to you once in a while”.
I don’t know about other bands but the bit about iron maiden is really stretched.
I guess if you consider the first lineup to be the one for their first concert in a bar’s basement, alright. But if you take the first album, Dave Murray was already in the band and still is.
And here it is. Of course a person like you isn’t gonna reread the thread and go “oh boy i kinda went overboard on this one and went straight for the jugular, i guess i was in a foul mood”.
Nah. Of course you’re gonna write a whole fanfic about how you’re actually trying to maximize good by… telling someone whose point you barely understand that they are “a recent example of the American education system” and “imitating our president someday”.
And you do that, not because you have poor impulse control and emotion management, but because really you are fighting the good fight and trying to spread a positive message to the masses.
What a fucking useless way to spend calories.
Abrasive comments like that, with no content whatsoever apart from insults, are really the lowest points of the Lemmy experience.
No willingness to share or accept any information, just pure sodium. Despicable really.
I mean there’s no hierarchy of blood sucking genocidal empires. They are all shit. Buying from one or the other, what fucking difference does it make ?
Pourtant il y a des gens qui prétendent que ça change leur vie/travail
Ils utilisent des modèles frontière qui ont des performances bien meilleures. Par exemple pour l’écriture c’est effectivement impossible sans une fenêtre de contexte conséquente. Déjà avec 50K tu peux obtenir des trucs pas mal. C’est pas du Tolstoi mais ça se lit et surtout ça reste cohérent, ça respecte les beats narratifs que tu as définis, ça fait pas dériver les personnages vers des versions génériques d’eux mêmes, etc…
Le côté agentique est utile aussi, pas pour faire de “vrais” agents mais parceque ce sont des modèles fine-tunés à bien tolérer les inférences longues et laborieuses. Ils restent dans leur voie, font des checkpoints réguliers pour se réaligner avec le contexte original, etc… En comparaison, les modèles de la génération précédente genre Llama et compagnie sont complètement arrachés. Le modèle va totalement vriller au bout de 5000 tokens et partir à 90° de sa tâche initiale.
En local si tu peux faire passer une quant de GLM-4.5-Air par exemple, tu vas avoir des résultats bien meilleurs. Bon ça va pas être particulièrement rapide par contre…
Cette semaine je réécoutais Death Throes of the Republic de Hardcore History. C’est fascinant de voir tous les parallèles qu’on peut faire avec l’époque actuelle ! Et puis c’est rocambolesque, l’histoire de Caius Marius et Sylla ça ferait une série Netflix. Les épisodes sont passés derrière le paywall depuis un moment mais euh… j’en dispose si ça intéresse quelqu’un. Go DM !
Sinon, Isaac Arthur a sorti un épisode rigolo sur les mathématiques extra-terrestres et j’avoue qu’il prend un angle auquel je n’avais jamais pensé. En gros on se dit que les maths c’est universel parceque si je montre un, deux ou trois objets à un alien il va forcément avoir une construction mentale de ce que 1, 2 et 3 représentent, et il va savoir les manipuler de sorte que 1 + 2 = 3 etc… Mais Isaac présente des contre-hypothèses : et si la vie existait sous forme de nuages de gaz par exemple, de sorte que tout est indénombrable ? Tu aurais une civilisation qui a beaucoup de maths pour désigner des proportions, ou des gradients, mais pas vraiment de concept de nombres entiers. Bref, épisode qui fait réfléchir dans une direction inhabituelle.
Reverse baiting is a thing though. I do it all the time on LinkedIn. Read a boomer’s post complaining about taxes, accuse them of being part of a whiny generation that doesn’t know anything about hard work. It’s hilarious.
Companies that have had their whole data on Google/MS servers for 20 years certainly don’t care for privacy the way you and I do. But they are certainly realizing that US providers are not the way to go. Baby steps I guess.
Oh yes. Because Merz builds data centers, and Der Leyen is known for making IT decisions at EU companies. They also happen to be the queens of private investors.
testing is doubting
You’re right of course but they’re not arguing in good faith. Wrong excel formulas in 2026? This person is either full of it, or they’re trying to prompt gpt-3.5 in ancient Aramaic.
Haha don’t know if i’m ai loving but i’m in my 40s if it helps. I was really using boomers as a slur to refer to dumb suits on LinkedIn, didn’t mean to offend.