>70% of AI revenue is coming from just two companies— OpenAI and Anthropic >therefore this is a bubble is this guy retarded? Why is it a problem that the companies with the best models are making the most money? The fact that he constantly depends on this point says it all.
>>62760008will you relaxyou want the price at IPO to be as low as possiblelet the normies think it's all a fad/scam/whateverthey literally do not matter
>>62760008it's part of the broader point that it's a bubble driven by the shady business tactics of these two companies rather than being the product of a broad organic trend. when these pillars erode (which has already begun to happen) then the bubble will pop.
>>62760008>are losing the most moneyftfy
>>62760029I think they’ll release a new model just before the IPO and people are gonna be pretty hyped. Idk how the prices will act in the first few months though. I’m not convinced that bears are willing to put their money where there mouth is. They’ve consistently been proven wrong over the last 4 years>>62760030>shady business tactics whatever that means. The more reasonable explanation is that when a tech’s progress depends on the concentration of tens and hundreds of billions of dollars, you cannot have multiple companies springing up everywhere making frontier models. > when these pillars erode (which has already begun to happen)more vague nonsense
The bubble is more about the fact that as of right now, they don’t really make anything or generate revenue. What sector can you think of that they have completely replaced already? But they’ve certainly borrowed and spent loads of money on property and making their own chips that they need a billion dollars each
>>62760008There's no way that the frontier models have long term value EXCEPT if they can achieve network effects that hold onto their user base.The open weight/open source models are catching up way too fast.As soon as the frontier companies try to raise prices to capture profits, corporations will bail from them and use cheaper models that are already 98% as good.
AI is here to make businesses more efficient. As a consumer you can see this clearly working by how cheap goods and services have become
>>62760089I think AI is some kind of deep state scamthey needed something to get out of the 2022 recession, so they went with AIthey worked the propaganda machinebut nothing is coming out of it other than a bunch of chatbotsand now they have no idea how to get out of it, without crashing the entire financial sysemlol
>>62760089> What sector can you think of that they have completely replaced already?they don’t need to “completely replace” any sector to make revenue. At this point virtually every white collar worker and student benefits from using AI. Programmers are disadvantaged if they refuse to use the best models. Mathematicians and scientists are becoming more reliant on AI for research. Then there are hundreds of millions of people simply using it for web search, or to explain ideas, or to generate ideas, etc. When you have a billion free users and you’re not bankrupt, then clearly the models are efficient and there’s potential for ad revenue. And yes they’re already making lots of revenue. Anthropic had $11.5B in Q2 and that number grew from just $4B in Q1, so Q3 is likely higher as well. >>62760095The average person would prefer to use the same model since it has all their old chats. And no, open models are not 98% effective. If that were the case then everyone would already be using them instead. They are a distillation of a distillation of internal models at OAI and Anthropic that are legitimately scary intelligent, as seen from the recent batch of hundreds of math proofs for example.
>>62760101these companies are all led by deep state figure headsAltman, Jensen, Musk, these are all just salesmen, that have little to do with the companies themselvesyou think they are smart, but they arent>The interview then took a bizarre turn, with Huang admitting that he lacks some surprisingly basic life skills himself.>“There are a lot of skills that don’t matter,” he said. “My first confession, I actually don’t know my address… one day I had to pump gas — it was a few years ago. They needed my ZIP code, and I panicked. I didn’t know my ZIP code. I don’t know my telephone number. I forget these things. I can live with it.”
>>62760115>anon discovers that geniuses are quirky and lack common senselol you don’t realize how you just exposed yourself as just another goy
>>62760008it sort of is a bubble, nvda is going to $50
>>62760098What the fuck are you talking about? What goods/services are cheaper due to Ai? That's not how business works, that efficiency (if it actually exists) goes straight to profits. The day of corporations passing savings onto consumers died along time ago.please, give an example if you can, because you sound insane to me right now.
>>62760089>they don’t really make anything or generate revenueThat's literally the 2 things they do. They make cutting edge technology and they make a lot of revenue.They just happen to spend significantly more money than they make in revenue and the cutting edge technology is never cutting edge for that long.
AI literally just makes shit up most of the time, confidently.>just trust it bro!fukcing porrs
>>62760008The IPOs are a scam because these companies can't stop losing money because models are a commodity. But what runs models, or their utility, that's no scam. That's where money is made.
>>62760008Just like how steam is a 'monopoly' on gaming
>>62760008>making the most money>making money
>>62760008
>>62760008>psyop of a gloweo
>>62760008>is this guy retarded?yesbut also>Why is it a problem that the companies with the best models are making the most money?it isn'tthe problem is the missing moat and overinflated expectation on usageAnthropic and OpenAI are just decent, open weights frontier model away from losing a big chunk of their paying customersNvidia and all Memory makers are just one improvement in LLM architecture away from losing their ludicrous profitsand long term success hinges on being able to undercut the $€元/intelligence of everyone elseliterally the shittiest market position imaginable, extreme competition and barely any lock in effectsas for most companies, changing from one to another model is a single line backend change
>>62763662> open weights frontier modelThere’s nothing preventing OAI and Anthropic from making cheaper and more intelligent models than everyone else. They have the best internal models and can distill before everyone else> Nvidia and all Memory makers are just one improvement in LLM architecture away from losing their ludicrous profitssure, if the Jevons paradox weren’t a reality. And on this note, there will always be demand for frontier models. No one will use shitty Chinese models when you can tell Claude or ChatGPT “make me a cool game” and it perfects it in one shot
>>62760110>open models are not 98% effective. If that were the case then everyone would already be using them instead.but they are already used everywhere anon!?and >>62760095 is absolutely righteven todayin actual industry adoption outside of coding and chatsopen (weight) models dominateeven when not being at the pareto-frontierbecause predictability, hosting location and access are more importantcurrently that doesn't really matterbecause the biggest paying customers are "developers" that prefer Claude/GPTbut as soon as this group is saturated, they will have a problem growing more
>>62763671>There’s nothing preventing OAI and Anthropic from making cheaper and more intelligent models than everyone elsethat assumes they have some kind of insurmountable leadand that progress will never stallwhich still wouldn't be a great positionbecause you constantly have to spend to stay aheadand as soon as you assume that others can catch upit's just a matter of time until there is an open frontier model>if the Jevons paradox weren’t a realitythe problem is that demand grew much faster then production capabilities take to scale upa memory fab takes 2-3years to build from conception to scale uptemporary efficient improvement are enough to alleviate the shortages and bring prices down (by a lot)>there will always be demand for frontier modelsI don't think anyone suggests otherwisethe question is moreare the high evaluations (and capex) justified for the expected frontier demand
>>62763692> that assumes they have some kind of insurmountable leadOAI and Anthropic have thousands of employees with currently superintelligent internal models, so their advantage is pretty strong. You can’t say the same for open labs.> and as soon as you assume that others can catch up it's just a matter of time until there is an open frontier modelEven if it existed, only those with hardware could use it, and this wouldn’t necessarily be cheaper to run due to economy of scale. >temporary efficient improvement are enough to alleviate the shortages and bring prices down (by a lot)That doesn’t follow at all. If you had access to Opus 5.5 at 100x efficiency there’s no reason not to use it. Nvidia would still sell all their chips, especially since they’re the most cost-performative at scale. You’re a moron.
>>62760008>He thinks AI models are making money You fucking retard, Sam Altman himself admitted to not knowing how he's going to properly make money from his AI model. It's all propped up to keep the numbers going.
>>62763739the actual inference is kinda profitable right now alreadyhosting providers that serve open models (almost) all make moneyAltman just doesn't know how to make their current modus operandi profitablethat includes the massive sums spend on training and research staffwhich they need to stay ahead of all the much smaller labs
>>62760008This is literally a fucking bait thread. OP selected a video infamous among redditors who follow ed zitron religiously for its ragebait potential, probably one of his most oppositional appearances where the host is a literal con artist moron who sells courses and who can barely speak or research his way out of a wet paper bag. Excellent psyop, truly. Bravo. OP is a salty SWE retard who probably got laid off and thinks he REALLY won't find a job if everyone runs for the door too early. Gotta keep this going as long as possible sucking hyperscaler cock so he can get another one. >>62763789You are a fucking moron. That bullshit you're talking about being "profitable" is precisely not profitable because of the training. The training not only needs to keep going to beat out "smaller labs," (what smaller labs, retard? its literally an anthropic and openai shell game, thats fucking it, everything else is larp peanuts using their same hardware and codex), it's also necessary to keep the models recent and up-to-date. If they stopped "training and research" 2 weeks ago the models would give you the news scoop from 2 weeks ago for the rest of time unless you started catching money on fire again. Not to mention the entire datacenter has to get replaced every time it wears out. Theoretically they have so many GPU's stored this is not a problem, unless you start thinking about how the longer they sit in a warehouse the more outdated they are. And they are out of date in a year or two and have about a three year shelf life when they are plugged in. Nvidia is lying to the government and banks telling them they have a 10 year shelf life while plugged in. LM FUCKING AOI keep getting forced to verify my email to use this board after I post AI scam facts. The jannies are banning people doing this, like they were banning people for shitting on the shitcoins they were shilling. I bet this thread is a part of that BS, honestly. More plausible than some retard SWE
>>62760008>Why is it a problem that the companies with the best models are making the most moneyThey're not making money, they're spending money. The fact that you don't understand the difference is why you shouldn't be investing.
>>62763789>Altman just doesn't know how to make their current modus operandi profitableAnd you don't think that's not the entire problem? A company that's worth litteral billions or dollars has no way of knowing how their business model will be profitable? How sustainable do you think this is?
>>62760064>shady business tactics>whatever that meansTHEY DONT MAKE MONEYTHEY'RE BUSINESS MODELS WITH NEGATIVE REVENUEEither they're totally lying about what they have in the pipeline, or they're lying about their mission. Based on public info, this is textbook scam
>>62763805what the fuck are you talking aboutthe expensive part is the pre-training which burns through millions of GPU hoursfine tuning and RL phases are much cheaperand architectural progress makes pre-training cheaper and cheaper (per weight)open weights (near) SOTA from scratch model pre-training usually costs in the region of a couple millions (7 digits)>keep the models recent and up-to-dateyour knowledge on this is >1y oldyou can feed newer information into the models with RAGs, Fine Tuning & PEFT or continual pre-trainingall are significantly cheaper than a new training runand every lab integrated all of them>entire datacenter has to get replaced every time it wears outyou're completely delusional about the timelinecapex break even period for large (open weight) LLM inference is at 3-6monthfor smaller models even shorterwith running cost included, you are in the green within <6month most of the timeGPUs with 80% util run a few years, maybe not 10, but doesn't matter
>>62763805A retarded Ed Wigtron cultist in the flesh lmao. How do you feel about the fact that Ed said LLM’s peaked in 2024? Or that the bubble would pop no later than Q2 2026? Or that Anthropic wouldn’t have $34B revenue in 2027 (they already passed $65 ARR in 2026)? Or, a month again when he said OpenAI wouldn’t have ARR of $60B in the next 3 months (they already hit $70B)? Or, when he was on the more recent DOAC podcast and asked “why aren’t you trying to slow down AI?” to the guy who literally wrote “If Anyone Builds It, Everyone Dies”? Or around the 39:40–40:15 mark of the first pod when Ed suggested inference is too expensive and when Steven simply asked “well do you know what the actual unsubsidized cost is?” Ed literally said “gosh it depends, now I wanna talk about something else”?Speaking of inference:> is precisely not profitable because of the trainingSo it seems you admit that inference itself is profitable but you think training costs won’t allow it. The problem is that updating models is NOT the same as testing out experimental algorithms, data sets, gigantic pre-trains, etc. We don’t even know how many failed strategies they’ve tried and failed. The models are only able to get better because they do giant batches of post-training with reinforcement learning, sometimes by letting the model itself generate hundreds of attempts at solving hard problems and selecting for the valid responses. All of this takes way more compute than simply adding updated data sets lol.
>>62760110>At this point virtually every white collar worker and student benefits from using AIThey all benefit from personal computers too.
>>62764037>>62764055>>62761938>>62761865>>62760057>they aren’t profiting now so they will never profitand goycattle wonder why theirs goycatfle. No really, take a hard look at yourself and your entire lineage. Ask yourself why your genes have never amounted to anything. You’ll find that it has very much to do with your misapplication of delayed gratification. Or maybe you do actually understand these things but for whatever reason really want AI to fail so you’re coping and turning yourself retarded to utter these points without realizing the cognitive dissonance.
>>62764048>And you don't think that's not the entire problem?no not reallyat least not by itself and not right nowthey are still in a massive growth phaseand there is no reason to chase profitability right nowat some point they eitherhave outgrown the problems, i.e. inference / product profits offset the training and research costorthey cut cost and expand revenue options>A company that's worth litteral billions or dollars has no way of knowing how their business model will be profitable? How sustainable do you think this is?you're confusing product with company profitabilityselling tokens / inference -> profitablestaying ahead of the competition with current spending -> unprofitableAnthropic and OAI both make a lot of money on their APIs and subscriptions
>>62764083>t. wall of copeI don't like Ed Zitron. I arrived at these conclusions naturally in 2022. My cousin was working in startups and came to me talking about AI is a conscious being and we burned a medicine wheel to the machine god at our tech retreat. These retards were building fucking point of sale systems. I also know how much a graphics card costs. Keep typing up that retarded bullshit though, faggot. I clicked on that video once because it kept getting recommended to me after I watched him on Bloomberg for 3 minutes. Bunch of redditors seething in the comments that the brown course scam artist was a course scam artist. How do you feel about being on the same side as a brown guy selling courses? Are you indian by chance? How is your H-1B visa going? Trump's supposed to be revoking them all. Thank fuck.
>>62764083>> Or, when he was on the more recent DOAC podcast and asked “why aren’t you trying to slow down AI?” to the guy who literally wrote “If Anyone Builds It, Everyone Dies”? That "author" is literally a Haredi jew pedophile who wrote a 660,000 word harry potter fanfiction where he self inserted himself as an 11 year old. He runs a fucking indian bay area rape orgy with a fucking street walker from Indiana. He did not graduate middle school. HAHAHAHA WOW LMFAO A RATIONALIST IN THE WILD! KILL YOURSELF! FAGGOTS LIKE YOU TOOK MY COUSIN'S MIND FROM ME AND NOW I HAVE TO HEAR ABOUT IT AT THANKSGIVING!
>>62764083>>62764101How are you calling people goycattle when you read the works of a haredi jew pedophile who dropped out of middle school, and maintains control of the bay area social scene by way of a literal eyes-wide-shut CNC orgy where they traffick young impressionable programmers to get raped by indians? Goycattle? You seriously need to get the fuck off this board. What are you doing on 4chan? Fucking region lock india
>>62764116> I arrived at these conclusions naturally in 2022I see, and your ego won’t allow you to change your mind in the face of new information. You say you don’t like Ed but called his DOAC podcast “one of his most oppositional appearances”…. about a guy who isn’t oppositional at all lmao. Clearly you’re a cultist who’s really mad that his daddy was interviewed by a guy who asked him basic questions like “what’s the actual cost” and Ed had no fucking clue how to respond. You do a really bad job at pretending not to be plebbitor lmao
>>62764136
>>62764136Saying "I see, and" is a turn of phrase not commonly found in the casual dialects of most North American native English speakers. More likely indian. >and goycattle wonder why theirs goycatfle. No really, take a hard look at yourself and your entire lineage.ESL> ...your misapplication of delayed gratification... utter these points without realizing the cognitive dissonance.You learned to speak English in a Madrassa in Pakistan
>>62764072I don't really give a flying fuck what open weight models are capable of, honestly. Their only use is setting up automatic inboxes that piss off your customers and coworkers and being integrated at companies on locally owned hardware so they don't have to permanently shill out millions to Anthropic anymore. The obvious downside of this is that those companies then have to maintain the servers and hardware themselves to run them, and are thus going to probably only build as much as they need for practical reasons (actually giving it to SWE's and cyber retards) but it eventually will be rationed out instead of tokenmaxxed and given to everyone and ultimately become another unremarkable part of a job that is functionally adult daycare for people with degrees in advanced economies who create them to satiate their populace and create a class of people that can act as a buffer against extremist social movements. You probably have one of those very jobs and probably think it is very important. It isn't. I'm sorry, buddy. We simply just don't need more code for the starbucks app. 99% of the shit running the entire planet was written by someone who's 10 jobs away and most faggot junior engineers who are hyped on this shit like you work at startups that are probably all going to disappear when VC pulls the money out after Anthropic and OpenAI fold. Enjoy being homeless, retard.
>>62764072>RAGs, Fine Tuning & PEFT or continual pre-trainingI literally don't need to know what any of these words mean because you learned them at a job that is fucking adult daycare and all anyone actually needs to realize about AI is 1. OpenAI, Anthropic, Meta, Tesla/SpaceX are all literally going to collapse under massive debt burdens that cannot functionally be bailed out by the government because there is literally not enough government debt on planet Earth able to be issued out to save them all forever, even if they started to wean off maintaining frontier models tomorrow and Trump utilizes the full economic weight of the western world to save them2. None of this will ever develop into anything resembling AGI, ever3. There is no usecase aside from software engineering, stealing solutions from a field that was useless and a non-contributor to the economy when ted kacynski started his terror spree decades ago, and call centers. And the amount of people working in call centers is only growing despite this bullshit tech being around for 6 years because people get mad and would rather talk to a humanThat's literally all I or anyone else on planet earth needs to know. The collapse of just Oracle alone will cause such a cataclysmic shock to western economies at large that millions of people like you and everyone you know and love will all be lining up at fucking food banks and running credit cards down to pay for rent. You better get ready for that. No country was ever supposed to have a third of its population employed in the tech industry. Hate to break it to you. Code just isn't worth that much money.
>>62764083>they already hit 70B
>>62764083>they already hit $70B>>62766257this aged really well
>>62764083>they already hit $70Bkekdid not age well
AI is a bubble because compute is a commodity. Simple as. Open models are the endgame of AI, race to the bottom (in cost)
>>62764243This guy gets it. I think the reason for the hype and literal terror tactics right now is that the scaling problem has already hit home - remember all the talking there used to be about scaling issues in the past? Suddenly noone remembers that, it's not because it's not happening it's because it happened way too soon. After all those billions they have nothing to show for it except chatgpt+ with some added refinements and schizo stories about how chatbots specifically trained to hack stuff hacked stuff. The real existential danger to humans is biotechnology, that's where real nasty things could happen due to messing with DNA and shit.
Debating whether AI is a bubble is completely irrelevant when it comes to making money off it.
>>62760098>AI is here to make businesses more efficient. >As a consumer you can see this clearly working by how cheap goods and services have becomeHAHAHAHAHAHA :)))) name ONE thing that's cheaper. Great bait though.
>>62763625This is great, anon. Libraries - the real data centers.
The play is to just invest in +30GB GPU and start runnning models locally and start replacing websites, video editors and other small services that people pay for. There are hundreds of sites that sell PDF conversions and other retardedly easy shit, that you can just take over
>>62760008>diary of a niggeryou're a nigger too for making this thread
>>62764243The US is literally going to pay off their debts you retard
>>62764202>>62764243This anon gets it correctly. Have a bump.
>>62760119Are you retarded
>>62764101absolutely seething and frothing at the mouth kabbalah tranny
>>62768301There's no way for the government to pay off their debts. They'd need to pay them a trillion dollars to bail them out every year and a half. You are retarded
>>62760008if that nigger is in your feed to even see that thumbnail, that says a lot about how retarded you are
>>62768564>There's no way for the government to pay off their debts.There is ONE way.Just, uhh, be sure to convert all your dollars into real assets before it happens.
>>62769662Right now, the only real hard assets are GPUs and memory. Gold, real estate, and Bitcoin are just memes.
>>62769665most of them sit unused in some warehouse somewhere
>>62760008i think ed has gotten a lot more level-headed over the years and has started to bring more receipts, but at the end of the day, he's still overly sensational with what he discusses, and the two biggest stories (if they are true) that he's brought up are the aging stockpiles of unplugged compute currently sitting in warehouses since a year and a half ago, and whatever the fuck is going on with oracle and whether they are truly strained. oracle is the more interesting of the two, since it's easier to look into purely from a financials perspective.rusckowski doing $3.5m in buybacks isn't a good look, though.there's no question that LLMs aren't useful, but everyone's already brought up the moat issue. LLMs are here to stay, but there is 100% financial fuckery going on with oracle that not a lot of people are discussing at all, and if you're resorting to relying on directors to perform buybacks on your behalf, something is rotten.
>>62769720there is no question that LLMs are* useful. as in, they are useful. don't know how i fucked that up but we
>>62769720> aging stockpiles of unplugged compute currently sitting in warehouses since a year and a half agoI’ve never seen any proof that this is real. I imagine this is just hallucinated from the fact that chips are bought before they’re actually created, or that they do exist but aren’t ready to be deployed in data centers, which says more about supply and less about demand. It’s also an exaggeration to say that they’re “aging” > and whatever the fuck is going on with oracle what? The force majeure that only applied to one (1) data center in New Mexico?
>>62769731This retard has been running this bait thread for 4 days now and stopped replying to me when I pointed out he was ESL and a student of eliezer yudkowsky lol>>62769665>>62766803These two posts are mutually exclusive. The GPU's can't be an inflation- or recession-proof asset if you're ambiguous on whether or not there is a bubble. You are concern trolling
>>62760098loud kek
>>62764101If what they said weren't true. Why do you personally feel the need to attack them this hard?
>>62769768I have a fetish for intellectually dominating people, especially the resentful failures in life
>>62769731>I’ve never seen any proof that this is realsame, which is why i don't find it interesting. it's also almost always directed at microsoft, and i can't find shit on it>what? The force majeure that only applied to one (1) data center in New Mexico?and their credit downgrade, yesterday's insider buyback that seemed reactionary to stabilize stock price, matured market penetration for its oldest products, 4% market share in the cloud space that makes their DC buildouts look like a hail mary into a space where google and meta have far more cash to burn than them, ellison's etchell, oracle and google couldn't be more different given the circumstances.if there's no model moat, and oracle has its eggs all in OAI's basket, what exactly is the story here besides "old, mature tech company with enigmatic billionaire at the helm bets that exactly one AI company's deal will quadruple revenue by 2031? " i won't say it's not possible, but the outlook and optics aren't good.
>>62769774
>>62769774>>62769779How many times you showered so far this year Aella?
>>62760064>when a tech’s progressthis sounds like determinism, thinking that just spending the money will automatically improve the outcome. >Mathematicians and scientists are becoming more reliant on AI for researchagain, assuming that research will amount to anything when quantum field phenomena is staring them in the face and functions outside of our current understandings of math. the same mistake, taking it for granted there's room for infinite growth>goycattle wonderah and there it is, another no social skills IT nerd looking to lord his technological superiority over us poor blue collar slobs
>>62769665Nice trolling.
>>62769720 >>62769720 >> aging stockpiles of unplugged compute currently sitting in warehouses since a year and a half ago>I’ve never seen any proof that this is realRecently . Oracle just canceled several of their warehouses, until they get power to them. Which could be 2 years out. Amazon just said recently that they take $8 bln of these chips and want some other company to have them on the balance. >why? because, obviously, they aren't worth 8 bln. https://www.youtube.com/watch?v=KIdL8g5hlWchttps://www.youtube.com/watch?v=YhiJGLCmOxshttps://www.youtube.com/watch?v=6FOsr_7_mVo https://www.youtube.com/watch?v=ZA1WOHKqcrg https://www.youtube.com/watch?v=UswBt8njrt4
>>62769731Replied to you here:>>62770400
>>62770400how has larry epstein not been luigi'd yet?
>>62770952that's how efficient totalitarian state with american characteristics really is
>>62760089>What sector can you think of that they have completely replaced already?It's about replacing workers and that is happening across the entirety of white collar work.They don't need to replace sectors, they just need to increase productivity while decreasing operational cost.
>>62760119>da eynvidya ceo is a genius
Ai doesn't exist. It can't even figure out that jews did 9/11 or produce a list of everyone that traveled to epstein island via flight logs. All it can do is fetch the state narrative and regurgitate it. Its a search engine pretending to be a redditor. Its absolutely fucking vile.
>>62760064>whatever that meanslollet the adults who work with finance post on this board, thanks. you can let your parents explain circular financing to you when you're older, Timmy.
>>62771680It's more expensive right now than hiring workers you fucking retard. The average faceless white collar worker outside of sales and tech makes like $50k a year. Even while it is being directly subsidized by the companies themselves by charging lower prices, direct government investment, and venture capital, individual costs to use frontier AI clear $10k per month on the low end for technical positions. If they were being honest with the price enough to make the company profitable as is today an individual ChatGPT Pro subscription would cost you thousands of dollars a month. This is not sustainable when an entry-level worker costs fucking $20 an hour and many will work part time because they are either shitheel rich kids who have their lives subsidized or often have two jobs well into their thirties. https://stocktwits.com/news-articles/markets/equity/meta-and-microsoft-reportedly-trim-anthropic-reliance-as-internal-ai-tools-take-center-stage/cZDqQVcRBjr The argument that it will "decrease operational cost" is not based on today's costs and mostly comes from people's emotions that it will just magically get cheaper. This is the same logic behind believing it will achieve RSI or AGI. It's your emotions. When a tech person says they're impressed with what Claude can do, it is an appeal to authority largely also based on their emotions.
>>62771857So the AI scam is a decentralized enron. Makes sense. And then you can't hold anyone liable since they were all acting independently and not coordinating the scam (lol)
>>62771857Why wouldn’t nvidia buy shares of OpenAI for example? If LLM’s succeed then Nvidia succeeds also. It’s not complicated lol. Another dumb cope like Ed’s argument in the OP that doesn’t withstand 5 seconds of actually thinking about it beyond the superficial vibes
