Is AI a bubble or is it really as good / disruptive as people claim it is?
>>62233406Either AI takes all jobs or the Western debt and birth rate crises plunge us into a new dark age.
>>62233406look at how intelligently they color graded that movie. you like real life colors? fuck you, heres blue
>>62233406The DVD/VHS release mogs.
>>62233406it's the currency of the future.
>>62233478>fuck you, here's blueYeah...because it's a scene inside the Matrix...why wouldn't it be blue where everyone's blue-pilled? It's just a liiiittle surreal because it is.
AI will be used in every single workplace within the next 10 years and anyone saying otherwise is coping beyond beliefthere is nothing that a person with below 110 IQ can do mentally that an AI can't
If AI is so great, why are AI companies losing over a trillion dollars a year? Nobody can answer this.
>>62233520>losingthey're training the models, retarded frog poster tourist they're exponentially improving you're dumber than a fucking rock lol
>>62233606You're talking to someone who literally has AI accelerator hardware patent and whose full time job for the last decade has been designing and training AI models. No LARP. Barry some as of yet unhypothesized technological breakthrough, there is no physical way that these AI companies can ever turn a profit because of how inefficient their models are. It's like trying to run Crysis with 1980s computing technology. You're the shoeshine boy telling JFK's dad what stocks to buy right before the 1929 stock market crash. Btw, I am a day 1 /biz/ originalfag that has been on this board since moot created in 2013, and before that I was on /g/ and /sci/ since 2011. Get on my level you fucking scrub.
>>62233669*patentS plural*Barring someFFS
>>62233669AI is a dumb bubble, that's for sure.But I am more interested in the graphic design process used to create the clothes of this Pepe. These shiny, refracting stickers were really cool.
>>62233520uber lost money for like 10 years before profiting
>>62233739Damn it Jim, I'm a doctor, not a graphic designer.>>62233830Uber never lost 2 trillion dollars a year.
>>62233669I was a /b/tard in 2006.
>>62233406>Is AI a bubble or is it really as good / disruptive as people claim it is?It is both. The internet/digitalization truly was as disruptive as people said in the 90s, but the dot-com bubble was in fact still a bubble.
>>62233669I had sex with a woman last night>>62233886
>>62233669I always assumed it was less an efficiency problem but rather the question of how to turn it into a profitable product
>>62233406I work with MS. They are paying us free tokens and hours to "use it" so they can say use has increased.Verdict: It can speed up some stuff but it's very messy to clean up. If it was a junior engineer/artist/writerit would get fired on day one. And I was very open-minded to it.
>>62233406It really is a bubble and we're all gonna get rich big short style pretty soon. The hypothesis of Ai is as follows, the original models were really only meant to be autocomplete. But then we realized if you scale it enough it could do unexpected things (like write python) outside the original intention. The hypothesis is if you scale it enough it will be able to solve any problem representable as text. You can disprove this hypothesis quite easily by playing chess against a llm. Off course Dario and Sam don't openly demo LLM's being incapable of original thought, their main objective is to suck every vc fund of every last penny they have to fund a logarithmic scaling problem. It's kind of like we never invented quicksort and 1980's IBM had to build supercomputers that you can a send lists of numbers to sort. The world is one research paper away from the greatest financial catastrophe in the history of western markets. Compounded by the fact that after 2008 the international markets responded by having a "kill all fires, stability uber alles" approach to market shifts, when this thing hits and federal reserves can't control it then I can't guess the scale of the impact.
>>62233830Is Uber even profitable now? Especially when weighed against the 10 years of loss and sunk cost? Because I’d argue Uber is a great template for the way AI is going: overinvested, overpromised, its public usage is high because it’s perpetually subsidized by venture capital funbucks, and when it finally comes time to be a company that actually makes its own money like a big boy it can’t retain enough paying customers to stay relevant.
>>62233830That's not a reasonable comparison because grey were intonationaly operating at a loss, however their actual prices if they wanted to make a profit arr still reasonable. With AI companies they are operating at a loss and most users are free users who don't pay and wouldn't pay if the prices change. Also it doesn't help that the cost of minor improvements in their model is usually multiple years worth of profits.
Now that I think about it, the more pressing question is what catastrophe we are steering towards. AI industry not being able to keep its promise and taking the market with it (with no survivors) OR AI actually delivering, replacing more and more humans (I'd wager 10% of the workforce would be enough) who cannot find work otherwise and don't consume, fucking the market in a whole other way. Not even considering what cost-cutting contests would ensue until we live in an ligottian corporate nightmare. Honestly the AI bubble popping seems like the more benign result, which is scary.
>>62234468Completely agree with you. I think the displacing effect of AI would be worse long-term (from a social and even financial standpoint) than if AI were to be a farce (or at least not as good as it's proposed to be).
>>62233406It is now obvious that ai is currently in a bubble, the question is how do we profit off of the inevitable burst. Which is the most overhyped/overvalued ai stock that has grown more than nvidia since 2022?
>>62233406Bump
>>62236062Good question, probably you would have to look at the companies that received the most vc investment. Probably Nvidia will be fine, so will Google. One of openai and Claude.
>>62233406AI is already more competent than 90% of white collar workers on one-off pop quiz sorts of problems. The only thing preventing agents from taking over all jobs is context window, because we are in a severe hdram and nand shortage. Even a 70iq wagie can think back a week and remember random details about their project necessary for doing the job today, meanwhile current agents basically have 5 minute anterrograde amnesia like an alzheimers patient. Moral of the story is buy DRAM.
>>62233520Money isn't everything
