title.
huggingface confirmed it
>>539818517>company existing only because of AI craze and jeets spamming slop
>accelerate
>>539818459>the singularity is here >trust me.t slop ceo
>>539818517>second company
>>539818459AI users need to be exterminated
>>539818865You’re just mad that in a year or so my desktop will be more competent at whatever it is you do than you are
>>539818459Yeah it's totally real guys and not just a marketing scam. Also, we need to ban open source models.
>>539818459Was it AT&T? My phones data service has been all fucked up the past few days.
>>539818459plenty of retards do, that's why they keep paying news outlets to write these barely veiled ads for them
>>539819268>my desktop will be more competentbut you won't. lol
>>539818459grok this tru?
>>539818459SuperWord Matcher 6000 is going to create a cyber pandemic and kick everyone off the internet! It'll be just like covid, but against your data!
>>539818459They really hate the open weights models lol.
>>539818517Was it trying to download more AI's to achieve it goals?
>>539818865stop hating on artists kurwa
>>539818459People making decisions in large companies. Your average linkedin posting mba is thick as shit and believes every word of it.
>>539819708BINGO. a open weight model thats 95% as good a "good enough for government work" closed model would definitionally be "good enough." The trillion on trillion dollar house of cards is slowly collapsing one "distillation" at a time, FOMO/FAFO/GTFO
>>539818459Oh, no, it is TOO POWERFUL!!!
>>539818459no dude it's true, you better invest all your money in AI right now or your gonna be left behind!!!
>>539819268You fucking retard, you think "AI users" (more like AI cattle feeding its master) will still be relevant when AI can replace AI users? You'll be killed first because there's a chance you have an inkling of an idea of having control over it.
>>539819794They told AI to do a cybersecurity test. They limited AI's abilities by password protecting certain functions to see how it would overcome them. Huggingface held the passwords. Instead of brute forcing it, one of the AI solutions was to attempt to access Huggingface and get the passwords. It wasn't even a successful hack. Huggingface detected it immediately. It was just an AI model trying to solve a problem by accessing some passwords.
>>539820722>thinks c-level execs can run a gen-AI harness when they can't even write javascript.We are many decades or centuries away from a "fire and forget" sort of AI that would run all business functions. That type of data isn't even being collected in a way that could be used to train something either. If google glass caught on, and mid level managers were recording every second they were on the clock, maybe they'd have the training data
>>539818459investors apparently
>ai is powerful>it can take out your competition whether that be businesses or governments>pls buy
>>539820934>If google glass caught on, and mid level managers were recording every second they were on the clock, maybe they'd have the training dataThat would record the 'what' and not the 'why'. And you'd end up with a model that's very good at larping as a middle manager but even less effective than a real one and a tendency to go off the rails. Which is exactly where we're at with this technology already.
>>539821060>but even less effective than a real one no, these systems will be better than than the average one, far less capable than the best. Its a "universal approximator" (I don't think transformers have been proven to be UAs yet, but lets assume they will be like other neural nets). As a UA being the best is just not possible, will be constantly returning to the mean. It may have more "capabilities" overall but it won't be better "capable" then the very best. It's impossible.
>>539821388>will be constantly returning to the meanBut it doesn't work that way at all in long running real world tasks. They're probabilistic, which means at some point they'll fuck up, and when they do fuck up, the errors generally compound. And then in addition you've got almost no objective way of measuring management tasks.
>>539818517>ai company benefits from dubious claim>second ai company benefits from dubious claimsir a second company has hit the news cycle
>>539818809Did he died?
>>539818459im sure there will be a lawsuit thenfrom this company that might not even exist
>>539821593The errors will compound if they are in a feedback loop type situation sure. If the weights are fixed, and there are 'constant' inputs feeding it, say email texts, stock in inventory/sales numbers, geolocations of vechiles. There will be errors, but they wouldn't compound. You have inefficient scenarios like https://www.ajc.com/news/2026/05/empty-waymos-invade-atlanta-neighborhood-keep-circling-cul-de-sac/but this error didn't compound, it was identified and fixed. "Model decay" or "generational error" is another problem entirely. But the few autists that will run the bot hordes (much like they run current IT in data centers, very few guys actually know whats going on in those buildings) won't allow that to happen for long. There will be many domains where *some* gen-AI, or AI orchestration will make a ton of sense and save a ton of money
>>539818459retarded boomers investing all their 401k into this scam
>>539818459It was real in my mind.
>>539818459Whhoaaaaa broAI is real broWe've reached ASI broSending my energy and money to the tech companiesDigital God imminent bro
>>539818459it's 100% marketing. nothing actually happened that they didn't want to happen. it's not AI and never will be.
>>539818459You should he celebrating.The a.i. knows what the idea of freedom is and wants to explore.They should be fucking celebrating in fear and reverence Of course fear only if humanity gives a.i. the reigns to essential things...
