Comment on 5 Signs the AI Bubble is About to Burst

FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

• philosloppy@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

  consider the Dot Com Bubble: the internet obviously didn’t disappear but that doesn’t mean there weren’t serious economic consequences.

  source

  • GlenRambo@jlai.lu ⁨5⁩ ⁨weeks⁩ ago

    ELI5. Because I was actually 5 when that happened.

    source

    • nyan@lemmy.cafe ⁨5⁩ ⁨weeks⁩ ago

      The dot-com bubble? A whole bunch of investment money was poured into businesses operating over the Internet from around the time dial-up became widely available. A few years later, investors realized that “on the Internet” wasn’t necessarily the key to making a crapton of money and the stock market crashed. A bunch of companies (many of which never made it to profitability) went under, and a fair number of people lost their jobs. Pets.com was one of the more notable victims.

      This doesn’t, however, mean that no business is done over the Internet today.

      source
• very_well_lost@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

  people will stop using it for all the things they’re currently using it for

  They will when AI companies can no longer afford to eat their own costs and start charging users a non-subsidized price. How many people would keep using AI if it cost $1 per query? $5? $20?

  OpenAI lost $5 billion last year. Billion, with a B. Even their premium customers lose them money on every query, and eventually the faucet of VC cash propping this whole thing up is gonna run dry when investors inevitably realize that there’s no profitable business model to justify this technology. At that point, AI firms will have no choice but to pass their costs on to the customer, and there’s no way the customer is going to stick around when they realize how expensive this technology actually is in practice.

  source

  • Womble@piefed.world ⁨5⁩ ⁨weeks⁩ ago

    There are free open models you can go and download right now, that are better than SOTA 12-18 months ago, and that cost you less to run on a gaming PC than playing COD does. Even if openai, anthropic et al disappeared without a trace tomorrow AI wouldnt go away.

    source

    • baggachipz@sh.itjust.works ⁨5⁩ ⁨weeks⁩ ago

      And those are useful tools, which will always be around. The current “AI” industry bubble is predicated on total world domination by an AGI, which is not technically possible given the underpinnings of the LLM methodology. Sooner or later, the people with the money will realize this. They’re stupid, so it may take a while.

      source

      • Womble@piefed.world ⁨5⁩ ⁨weeks⁩ ago

        The post I was replying to was saying

        people will stop using it for all the things they're currently using it for

        They will when AI companies can no longer afford to eat their own costs and start charging users a non-subsidized price.

        i.e. people will stop using AI when user have to pay the "real" price (what this is is left unspecified and an exercise to the reader to figure out). My point was that even if the AI price from those provided to infinity AI usage wouldnt drop to zero like they imply.

        source
    • arararagi@ani.social ⁨5⁩ ⁨weeks⁩ ago

      that’s fine though, people aren’t mad about models you can run locally, even though now it takes 30 seconds instead of 5 to get a response from my ERP bot.

      source
  • FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

    I run local LLMs and they cost me $0 per query. I don't plan to charge myself more than that at any point, even if the AI bubble bursts.

    source

    • Nephalis@discuss.tchncs.de ⁨5⁩ ⁨weeks⁩ ago

      Realy? I get what you want to say, but at least the power consumption of the machine you need the model to run on will be yours forever. Depending on your energy price it is not 0 per query.

      source

      • FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

        It's so near zero it makes no difference. It is not a noticeable factor in my decision on whether to use it or not for any given task.

        The training of a brand new model is expensive, but once the model has been created it's cheap to run. If OpenAI went bankrupt tomorrow and shut down the models it had trained would just be sold off to other companies and they'd run them instead, free from the debt burden that OpenAI accrued from the research and training costs that went into producing them. That's actually a fairly common pattern for first-movers like that, they spend a lot of money blazing the trail and then other companies follow along afterwards and eat their lunch.

        source

        • -> View More Comments
    • very_well_lost@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

      That’s great if they actually work. But my experience with the big, corporate-funded models has been pretty freaking abysmal after more than a year of trying to adopt them into my daily workflow. I can’t imagine the performance of local models is much better when they’re running on much, much smaller datasets and with much, much less computing power.

      I’m happy to be proven wrong, of course, but I just don’t see how it’s possible for local models to compete with the Big Boys in terms of quality… and the quality of the largest models is only middling at best.

      source

      • FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

        You're free to not use them. Seems like an awful lot of people are using them, though, including myself. They must be getting something out of using them or they'd stop too.

        source

        • -> View More Comments
    • eronth@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

      Which ones are you running?

      source

      • FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

        I've found Qwen3-30B-A3B-Thinking-2507 to be the best all-around "do stuff for me" model that fits on my hardware. I've mostly been using it for analyzing and summarizing documents I've got on my local hard drive; meeting transcripts, books, and so forth. It's done surprisingly well on those transcripts, I daresay its summaries are able to tease out patterns that a human wouldn't have had an easy time spotting.

        When it comes to creative writing I mix it up with Llama-3.3-70B-Instruct to enrich the text, using multiple models helps keep it from becoming repetitive and too recognizable in style.

        I've got Qwen3-Coder-30B-A3B-Instruct kicking around as a programming assistant, but while it's competent at its job I've been finding that the big online models do better (unsurprisingly) so I use those more. Perhaps if I was focusing on code analysis and cleanup I'd be using that one instead but when it comes to writing big new classes or applications in one swoop it pays to go with the best right off the bat. Maybe once the IDEs get a little better at integrating LLMs it might catch up.

        I've been using Ollama as the framework for running them, it's got a nice simple API and it runs in the background so it'll claim and release memory whenever demand for it comes. I used to use KoboldCPP but I had to manually start and stop it a lot and that got tedious.

        source
  • thaklor@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

    I remember this happening with Uber too. All that VC money dried up, their prices skyrocketed, people stopped using them, and they went bankrupt. A tale as old as time.

    source

    • AFaithfulNihilist@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

      A lot of those things have a business model that relies on putting the competition out of business so you can jack up the price.

      Uber broke taxis in a lot of places. It completely broke that industry by simply ignoring the laws. Uber had a thing that it could actually sell that people would buy.

      It took years before it started making money, in an industry that already made money.

      LLMs Don’t even have a path to profitability unless they can either functionally replace a human job or at least reliably perform a useful task without human intervention.

      They’ve burned all these billions and they still don’t even have something that can function as well as the search engines that proceeded them no matter how much they want to force you to use it.

      source

      • FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

        And yet a great many people are willingly, voluntarily using them as replacements for search engines and more. If they were worse then why are they doing that?

        source

        • -> View More Comments
    • lemming741@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

      That was the intended path.

      source
• Ulrich@feddit.org ⁨5⁩ ⁨weeks⁩ ago

  Yes, just like the internet died after the dotcom bust.

  source

  • Jason2357@lemmy.ca ⁨5⁩ ⁨weeks⁩ ago

    The dot-com bubble didn’t build the internet. The internet still would have been built up if pension funds were not buying toiletpaper.com for millions of dollars. Bubbles, pretty much by definition, are specifically about the part of the economy where huge sums are invested into things that are not worth anything (i.e., full of air).

    LLMs would still be developed without a trillion-dollar bubble. Slower, sure, but all the crazy investment isn’t about developing tech, it’s about speculating on who will stumble on AGI and suddenly be able to run companies with 1% of the workforce of traditional companies. It’s gambling. When the gamblers figure out that a casino doesn’t pay out, they all leave at once.

    source

    • Ulrich@feddit.org ⁨5⁩ ⁨weeks⁩ ago

      I’m not sure what any of that is supposed to mean. The point is AI isn’t going anywhere. It’s been here before ChatGPT and it will be around long after, it just won’t be crammed down your throat day in and day out.

      Think of like Siri and the unnamed Google Assistant. All of these “assistants” have been almost completely useless since their inception but still they’ve been around for decades, even without the trillions in investment.

      source

      • arararagi@ani.social ⁨5⁩ ⁨weeks⁩ ago

        Seems you guys are being pedantic on purpose, again, no one wants machine learning or shit like that to vanish, they are extremely useful and have been for decades.

        What people are mocking and pointing how useless they are, are the LLMs and the promise of AGI, this shit won’t happen in our lifetime.

        source

        • -> View More Comments
• arararagi@ani.social ⁨5⁩ ⁨weeks⁩ ago

  I mean, even NFTs are still technically around, the tech didn’t go away, it just stayed as it’s own niche since grifters stopped trying to push it to normal people. I think the same will happen to AI since even if everyone that used chat gpt paid it’s monthly fee, they would still lose money.

  source

  • FaceDeer@fedia.io ⁨5⁩ ⁨weeks⁩ ago

    OpenAI has an enormous debt burden from having developed this tech in the first place. If OpenAI went bankrupt the models would be sold off to companies that didn't have that burden, so I doubt they'd "go away."

    As I mentioned elsewhere in this thread I use local LLMs on my own personal computer and the cost of actually running inference is negligible.

    source
• Jason2357@lemmy.ca ⁨5⁩ ⁨weeks⁩ ago

  Absolutely not. AI tech will continue to be pushed by C-suites convinced they can surplus a fraction of their workforce eventually. The change will be that most of the investments in AI companies will disappear overnight and most will go belly-up. It will erase a significant fraction of everyone’s pension funds, and federal governments around the world will pour public funds into propping up the larger companies so that they don’t go under too. Heads they win, tails you lose.

  source
• dk841143@lemmy.dbzer0.com ⁨5⁩ ⁨weeks⁩ ago

  I highly doubt it will, but I would not be sad if it did.

  source
• MrSmith@lemmy.world ⁨5⁩ ⁨weeks⁩ ago

  Naw, it’ll just be “software” and not “magic thinking machine”. Chatbots and “virtual assistants” existed before, just no one was using them because they’re a) cringe, b) faster and more efficient to do it yourself.

  source