11 ms·

The money situation is what gives me pause with LLMs. The amount of money it's burned on this is giant; those companies will need to make so much money to have

• Mistletoe 1y ago
  I’m constantly confused why Google wants me to use Gemini so much. Why would they want me to use an AI that doesn’t show ads, isn’t their search, and costs them money to generate the responses? The instant we see ads in LLMs we will stop using them, it will be as abrupt and as annoying as the vuvuzela example because using an LLM is a very intimate interaction. It will feel much more jarring when your intimate chat friend becomes a used car salesman to you. I get that AI=stock go up right now but it seems like they are poisoning their own well deeply down this path.
  • aleph_minus_one 1y ago
    > I’m constantly confused why Google wants me to use Gemini so much. Why would they want me to use an AI that doesn’t show ads, isn’t their search, and costs them money to generate the responses? The instant we see ads in LLMs we will stop using them, it will be as abrupt and as annoying as the vuvuzela example because using an LLM is a very intimate interaction. I think with the last sub-clause (that I emphasized), you answered your question: because the conversation is more intimate, Google learns more about your "true interests", be it to make advertising to you more targeted, or for more sinister purposes.
  • lipowitz 1y ago
    Google thinks it is going to charge a monthly fee for Gemini so they are basically correcting their model to be user paid in the switch from search to LLM. They are also trying to reduce workers in search, etc, somewhat through the trained LLMs so the other idea is that they have lower costs per user. I don't think many people will pay monthly fees or will want to pay them for each platform they use which is why they all tend to do so many questionable integration attempts to try to get users to not want to use a separate LLM of their choice in a browser instead.
    • k__ 1y ago
      This. Gemini Pro requires a monthly subscription. Seems like a pretty straightforward business model.
      • Mistletoe 1y ago
        How many would really pay though? I certainly wouldn't. Would you pay for Google search?
        • lipowitz 1y ago
          Their ads are very convincing, I'd pay up to 40¢ a month for Gemini to tell me how to recover with a recipe for tomato sauce I accidentally put sugar in, though only in months where I inexplicably add sugar to things.
  • pbhjpbhj 1y ago
    If you're going to use an LLM instead of their search better to keep you on their estate. Presumably they're using responses to tune a personalisation layer, or in some way, they can use to modify their advertising algos?
  • fragmede 1y ago
    > The instant we see ads in LLMs we will stop using them You mean like how the instant we see ads on YouTube we will stop using it?
    • sundarurfriend 1y ago
      Or like how the instant Netflix disables password sharing, we will stop subscribing to it.
  • dragonwriter 1y ago
    > The instant we see ads in LLMs we will stop using them But what if you don't see distinct ads? LLM advertising can just be paid-for bias in the generated content.
    • lblume 1y ago
      Moreover, the bias could be arbitrarily subtle. This is something that really worries me about black box AI systems.
      • MrGilbert 1y ago
        I‘d argue that, at some point, this behavior will surface enough that it gets interesting for government to take a look at it. At least in my country, ads need to be labeled as such - I don’t think this is going to change.
        • Lu2025 1y ago
          The entire business model of tech is to fly under the radar of regulations. They can't even force the bastards to stop tormenting us with cookie pop-ups, i seriously doubt the government will do much.
      • bjt 1y ago
        What marketing department on earth is going to pay for an ad campaign so subtle that they can't even tell if it's running? Ad budgets aren't bottomless. People making decisions about them have a lot of options for where to spend. They want provable attribution so they can tell which channel is giving them the most bang for the buck. If that exists, then the ads will be discernible.
  • ZeroGravitas 1y ago
    I saw my first AI ad yesterday, in Microsoft Copilot, recommending a Jetbrains product.
  • 9dev 1y ago
    Google has this pesky problem of carrying almost all their eggs in one basket: The moment their ad business doesn’t work anymore, the behemoth that Google is will come down to its knees. But due to the way the company is structured, they need search to drive users to the ads, and it’s all one big mess of entangled revenue streams that can’t be touched for fear of breaking something. Now if AI turns out to be the next big thing, they can steer differently next time, sell subscriptions, and avoid all that entanglement with multi-sided markets and layered revenue strategies. At least that’s my take.
    • h2zizzle 1y ago
      >The moment their ad business doesn’t work anymore, the behemoth that Google is will come down to its knees. Which makes it all the weirder that they seem to be intentionally sabotaging it by nerfing Search into unusability.
      • Marazan 1y ago
        The slow destruction of Google search is and will continue to be revenue positive for Google right up until the point where it isn't. Eventually they will break the "trust thermocline" in their search results and that will blow it up but on the way they'll keep making more and more money from every damaging change they make.
        • woooooo 1y ago
          Crucially, the people driving each incremental change get to celebrate the extra revenue they drove and get promoted for it.
    • autoexec 1y ago
      > they need search to drive users to the ads Why? The websites we visit can still be infested with google's ads, and so can our gmail accounts, and so can the youtube videos we watch, and they can push ads directly onto our cell phones 24/7. Google has plenty of ways to force ads into your life. Google used to need search in order to build extensive dossiers on everyone. It told them what people were looking for online. What they were interested in. Now Google has their cell phones, their browser, and their DNS servers doing that for them. Most people are handing all of their browsing history to Google. Google doesn't need search, which is why it's been allowed to atrophy into uselessness.
  • dismalaf 1y ago
    Because Gemini has paid tiers. Also to prevent users from going to other providers.
  • Xenoamorphous 1y ago
    > The instant we see ads in LLMs we will stop using them People didn’t stop using Google Search, Facebook, Instagram, YouTube, TikTok and a myriad other services and products (and things like TV before those) because they got ads.
    • tartoran 1y ago
      It's not just ads but plenty of dark patterns. At some point as a user you no longer have much control and no longer get what you were looking for and decide to leave. But the slow boiling leaves a lot of frogs in the pan, it will probably be a slow decay before death. To me browsing the net is unusable without ad blockers but there are many users out there who don't know what ad blockers are.
• closewith 1y ago
  > The idea is that we will all spend more money on AI that we spend on phones, and we will spend it on those companies only... I don't know, it just doesn't add up. My companies currently spend more on AI than phones - hardware and subscriptions. It's now the second highest expense to salaries and director's remuneration.
  • Mizza 1y ago
    What are you doing with it?
    • multjoy 1y ago
      Anything that could possibly justify the expense, I suspect, whether it is the right thing or not.
      • closewith 1y ago
        Isn't that true of all software development? But we are lily-white both legally and ethically. One of the perks to a lifestyle business beholden to no investors.
    • closewith 1y ago
      We make compliance software for EU enterprises that helps them remain GDPR compliant when writing software. We use agentic LLMs in our client-facing software development, which is largely industry-specific CRUD apps that use our existing APIs.
• immibis 1y ago
  More hopeful explanation: they'll never make their money back and this is either part of the great rebalancing or great collapse of the economic forces that be.
• sheiyei 1y ago
  <AIs are much better at responding to my intent, and they rarely attempt to sell me anything> YET.
  • jiveturkey 1y ago
    I would wager that the most prevalent use of AI today is to sell you ads. Whether through market analysis, campaign analysis, content optimization, and content generation.
  • pcthrowaway 1y ago
    Remember, Google also didn't have ads interspersed with their search results for over 1̶2̶ 2 years
    • kristianc 1y ago
      That wasn’t out of benevolence, that’s because they hadn’t discovered the ads business model yet. The genie is well and truly out of the bottle now.
      • wizzwizz4 1y ago
        They're on record as talking about how Google has better results because it doesn't have ads.
        • passwordoops 1y ago
          At that time and it's what got them the market share. Once they achieved monopoly status "Don't be Evil" was quietly replaced by "You call that an acceptable margin?!"
      • schmidtleonard 1y ago
        They always knew. > The goals of the advertising business model do not always correspond to providing quality search to users. - Sergey Brin and Lawrence Page, The Anatomy of a Large-Scale Hypertextual Web Search Engine http://infolab.stanford.edu/~backrub/google.html http://infolab.stanford.edu/~backrub/google.html
• bondarchuk 1y ago
  "It has cost a lot of money and therefore they will try very hard to make it back". I don't think it really makes sense. Companies will always try to make the maximum amount of money anyway.
  • bell-cot 1y ago
    Obviously true. OTOH, there are far too many people who desperately want to believe in cool new stuff really being free, without any "gotcha" down the line.
  • h2zizzle 1y ago
    No? Some companies are mission-driven. Some are ran by people with ethical scruples, and/or who will forego short-term profit for long-term brand image. Some (allegedly) have been infiltrated or taken over by bad actors who purposely sabotage the company's profitability. Some are private and simply do as they wish. I wish we wouldn't mindlessly repeat these platitudes. Try and falsify your statements before posting.
    • LeifCarrotson 1y ago
      Some are or at least start that way, but they exist in a competitive market. In our current regulatory and economic environment, it appears that mission-driven, long-term oriented, ethical companies are typically out-competed by finance-driven, short-term oriented, greedy companies. The article describes the struggle of using a search engine in 2025. Which is to say, using Google in 2025. Search engines benefit greatly from huge economies of scale, and most websites are optimized for Google SEO and for their ad network. Sure, the folks at DuckDuckGo (my search engine) or Kagi appear to be your good sort of company, but the revenue and popularity of those companies is a rounding error in comparison to Alphabet, Inc. They can't afford the crawlers and infrastructure of the big finance-oriented players, they can't convince most websites to optimize for their engine, and most people don't even know they exist. Sure, there's a handful of people running the equivalent of a small-town grocery with local farm-sourced produce and hand-selected general goods as a passion project, working long hours and slowly chewing through their savings. And there are a handful of people who feel that the existence of such a place is important, and shop there out of principles in spite of the incentives and penalties associated with that behavior. But most of the country is overrun by Dollar Generals and Wal Marts.
• Lu2025 1y ago
  Regarding money, they label a lot of expenses as R&D and write them off taxes. Taxpayers foot the bill to some extent.
  • andruby 1y ago
    That's not how accounting works. Companies pay salaries and pay for hardware. They don't make profit so they don't pay taxes. By labeling the salaries as R&D assets and amortizing that over 5 years (instead of taking it completely in the first year), they're more likely to make "accounting" profit and pay taxes in the early years. Those legislative changes will likely move forward the taxes being paid. But to your point: not paying taxes because a company is investing doesn't mean taxpayers are footing the bill. It does mean the company isn't contributing to paying taxes while it is in "growth investing" mode.
• jonplackett 1y ago
  They’re all investing with the assumption they can be the ‘winner’ and take all the spoils. Maybe nvidia can be a winner selling shovels but it seems like everyone else will just be fighting each other in the massive pit they dug.
  • philipwhiuk 1y ago
    NVIDIA is already a winner selling shovels. They don't need a winner, they want the race to continue as long as possible.
• Xss3 1y ago
  Ai is in its early youtube phase. Everyone loves that it's free and adfree and that its algorithm's primary purpose is to serve relevant content not profitable content, everyone knows it can't stay that way, and we are all waiting for the enshittification to kick in on the march to profitability. The question is, will AI chat or search ever be profitable? What enshittification will happen on that road? Will AIs be interrupting conversations to espouse their love of nordvpn or raid shadow legends?
  • tartoran 1y ago
    I'm aware this is going to happen. but you don't think offline solutions will be more prevalent by the time openai jacks up the costs? These companies have no real moats unless they start doing something social so they have a network of captive audience or something like that.
  • yuck39 1y ago
    Many of the traditional SEO players are now figuring out how to game the system to get their customers to show up more frequently in LLM responses. Once the pressure to turn a profit is high enough the big players surely won't just leave that money on the table. The scary part is that even if we end up paying for "ad-free" LLM services how do we really know if it is ad-free? Traditional services are (usually) pretty clear on what is an ad and what isn't. I wouldn't necessarily know if raid shadow legends really is the greatest game of all time or if the model had been tuned to say that it is.
• Cthulhu_ 1y ago
  What I'm seeing at work now is the companies selling AI stuff to companies - mostly Microsoft in my neck of the woods. For me as an employee, using copilot is a trivial thing, not my wallet. But just like with AWS where a developer doesn't really need to worry about how many machines their merge request starts up and lets run, the bills will start to creep up to the companies. For now Copilot is a fixed $20 / month / person, but it's only a matter of time before it becomes metered, or the advanced models cost more credits. This is also why they're pushing for agents, because a single query is cool and all, how much it costs in compute is reasonably predictable, but an agent can do a lot of interesting things and do 100x the usage of a single query, and put 100x the charge on the corporate credit card. It'll probably have a chilling effect, with companies being like "ok maybe let's tone down a bit on the AI usage", just like how they hire consultants to bring down their runaway AWS costs.
• jraph 1y ago
  > Maybe there IS such a thing as a free lunch after all! A free lunch that costs our environment though, which is a big caveat :-)
  • tartoran 1y ago
    The free lunch creates a lot of dependency on AI so when the lunch isn't free anymore it will bite hard.
    • Ferret7446 1y ago
      Same as all technology. You missed the deadline to file that complaint by more than a couple hundred thousand years.
  • scrollaway 1y ago
    Maybe. But to be honest, optimizing monstrously slow processes that cost weeks of human labour by automating them, that saves a ton of energy as well. It’s not zero sum, as the humans spend that energy elsewhere, but ideally they spend it on more productive things. This calculus can very quickly offset whatever energy is wasted generating cartoon images of vuvuzelas.
    • jraph 1y ago
      > optimizing monstrously slow processes that cost weeks of human labour by automating them, that saves a ton of energy as well Yes, I do agree with this. However, that's only good as long as there wasn't a better way of optimizing them. Assuming we'd not be better off getting rid of those costly process altogether. > ideally they spend it on more productive things Same gotcha as mentioned in my other comment: "productive" in our growth economy often means "damaging to the environment", because we are collectively spending a lot of our time producing garbage and that's not something we should really optimize. Most of us work a fixed amount of hours so it's not like we are doing ourselves any favor by optimizing time in the end. In another system, I wouldn't say. I'm generally for freeing up time for us so we can have better lives.
    • 1y ago
• usrusr 1y ago
  There are many phones that won't be bought when businesses can substitute some of their workforce with LLM. That market is not about B2C at all.
• marcosdumay 1y ago
  Yes, the question everybody awake is asking is how long until all the LLM corporate initiatives die? Because as useful as those things can be, they just can't do enough to justify that cost. But there are free (to copy) ones, and smaller ones. And while those were built from the large, expensive models, it's not clear if people won't find a way to keep them sustainable. We have at minimum gained a huge body of knowledge on "how to talk like people" that will stay there forever for researchers to use.
  • troyvit 1y ago
    > We have at minimum gained a huge body of knowledge on "how to talk like people" that will stay there forever for researchers to use. This is spot on. I think we'll be able to capitalize on other talents of "AI" once we recognize the big shift is done happening. It's like five years after the Louisiana Purchase: we have a bunch of new resources but we've barely catalogued them, let alone begun to exploit them. > how long until all the LLM corporate initiatives die? Sooner than I personally thought, and I place a lot of that with Apple's. They've led the way in hardware that supports LLMs, and I believe (hope?) they'll eventually wipe out most hosted chat-based products, leaving the corporate players to build APIs and embedded products for search, tech support, images, etc. The massive amounts of capital going into OpenAI, Anthropic, etc., will ebb as consumer demand falls. I hope for this because the question I keep asking is, how can our energy infrastructure sustain the huge demand AI companies have without pushing us even further into a climate catastrophe?
    • chasd00 1y ago
      > This is spot on. I think we'll be able to capitalize on other talents of "AI" once we recognize the big shift is done happening. It's like five years after the Louisiana Purchase: we have a bunch of new resources but we've barely catalogued them, let alone begun to exploit them. one thing about LLMs used as a replacement for search is they have to be continually retrained or else they become stale. Lets say a hard recession hits and all the AI companies go out of business but we're left with all these models on huggingface that can still be used. Then, a new programming language hits the scene and it's a massive hit, how will LLMs be able to autocomplete and add dependencies for a language they've never seen before? Maybe an analogy could be asking an LLM to translate a written language you make up on the spot to English/other language.
• autoexec 1y ago
  > As a user it's a great free ride though. Maybe there IS such a thing as a free lunch after all! if you consider the massive environmental harm AI has and continues to cause, the people whose work has been stolen to create it, the impacts on workers and salaries, and the abuses AI enables that free lunch starts looking more expensive.
  • baggy_trough 1y ago
    What is this massive environmental harm? That sounds like hyperbole.
    • ksenzee 1y ago
      They’re restarting coal-fired power plants to run AI datacenters. I don’t know what your personal threshold is for “massive” environmental harm, but that meets mine.
      • baggy_trough 1y ago
        What's a specific example of that?
        • svnt 1y ago
          It has been widely reported. Here is an example not of a coal-fired but diesel and gas mobile power sources. If you spend time looking you will have no trouble finding sources. > Last week, the Environmental Protection Agency (EPA) issued a rule clarification allowing the use of some mobile gas and diesel power sources for data centers. In a statement accompanying the rule, EPA Administrator Lee Zeldin claimed that the Biden administration's focus on addressing climate change had hampered AI development. > "The Trump administration is taking action to rectify the previous administration's actions to weaken the reliability of the electricity grid and our ability to maintain our leadership on artificial intelligence," Zeldin said. "This is the first, and certainly not the last step, and I look forward to continue working with artificial intelligence and data center companies and utilities to resolve any outstanding challenges and make the U.S. the AI capital of the world." https://www.newsweek.com/ai-race-fossil-powered-generators-are-data-centers-dirty-secret-2068791 https://www.newsweek.com/ai-race-fossil-powered-generators-a...
• crazygringo 1y ago
  > The amount of money it's burned on this is giant It's big, but it's honestly not that big. Most importantly, costs will quickly come down as we realize the limits of the models, the algorithms are optimized and even more-dedicated hardware is built. There's no reason to think it isn't sustainable, it will add up just fine. But yes, it will attract a ton of advertising, the same curve every service goes through, like Google Search, YouTube, Amazon, etc. Still, just like Google and Amazon (subtly) label sponsored results, I expect LLM's to do the same. I don't think ads will be built into the main replies, because people will quickly lose trust in the results. Rather they'll be fed into a separate prompt that runs alongside the main text, or interrupts it, the way ads currently do, and with little labels indicating paid content. But the ads will likely be LLM-generated.
  • ToucanLoucan 1y ago
    > the same curve every service goes through This is honestly why I struggle to get excited for anything in our industry anymore. Whatever it is it just becomes yet another fucking vector for ad people to shove yet more disposable shit in front of me and jingle it like car keys to see if I'll pull out a credit card. The exception being the Steam Deck, though one could argue it's just a massive loss-leader for Steam itself and thus game sales (though I don't think that would hold up to scrutiny, it's pretty costly and it's not like Valve was hurting for business but anyway) but yeah. LLMs will absolutely do the exact same, and Google's now fully given up on making search even decent, replacing it with shit AI nobody asked for that will do product placements any day now, I would bet a LOT of money on it.
  • exceptione 1y ago
    > I don't think ads will be built into the main replies, > because people will quickly lose trust in the results. The 'best' ads will be those the public doesn't recognize. Surf the internet without an ad blocker and you will die from a heart attack. This is a matter of conditioning users. It will take some time. Case in point: people already give up on privacy because "Google knows about everything already", which reflects a normalization of abuse, as we started from trust and norms ("don't be evil"). So, can they? yes. Will they? yes.
  • 1y ago
• babypuncher 1y ago
  We went through this with all the Silicon Valley "disruptors" in the '00s and '10s. It's fun while they're focused solely on burning VC money to build a massive userbase, but as soon as they decide it's time to start actually making money the deal gets a lot less sweet. Only by that time, most of the competitors are gone and you don't have much choice.
• adverbly 1y ago
  > The idea is that we will all spend more money on AI that we spend on phones AI is not a consumer product. Businesses will pay for AI. They will use it for whatever they are building. We will buy what those businesses build. The medium term here is that AI is going to become part of the value chain. It's gonna be like stripe or insurance or labor.
• thih9 1y ago
  Speculation: AI companies want the cultural shift, i.e. get everyone used to having their data, art, work, etc, turned into models. Plus they want the PR, i.e. AI agents to be seen as helpful, friendly and genuinely useful. They want this to happen fast and before legislators react too. Releasing for free seems safe and efficient.
  • ASalazarMX 1y ago
    And this is already happening. I have a few acquaintances that will ask ChatGPT thinks they could readily find on Wikipedia, or use it as a shortcut for mental effort. They've become dependent, even addicted, to this patient butler that can work and think for them; who cares if it makes some mistakes? They'll just ask it more. Once enough people become like this, they will gladly pay to keep it, they'll consider it a basic necessity.
• spaceguillotine 1y ago
  this is a perfect example of the VC churn, they hope to get unattainable and unsustainable results and don't care what happens to anyone that was doing it sustainably before they came around. The internet has become total garbage now all because a few men wanted to make a bunch of money by making silicon do the thinking for them. AI is the quickest route to ruin and ending up with humans like in Idiocracy devoid of critical thinking, and the output of LLMs is so bad to read, students are just turning in the worst papers using LLMs and learning nothing. At first my school banned use of them but then Microsoft tipped their hand because they donate a lot of money and now everyone is allowed to use AI and they got rid of the requirement to use MLA Citations so everything turned into slop.
• strangattractor 1y ago
  > As a user it's a great free ride though. Maybe there IS such a thing as a free lunch after all! I remember when search was a free ride. The articles that I found in searches where relevant and there was no wordy boiler plate AI content specifically designed to get me to see all the advertising on the page. There is no free ride - AI will accelerate the enshittification of the Web by orders of magnitude. Barriers to garbage content generation are rapidly approaching 0.
• Exoristos 1y ago
  Doesn't it seem suspect that a product with such massive investment is made available for little or no cost? Either it's garbage (you don't get AutoCAD for free: you get meme generators), and investors are getting soaked; or it's digital gold, and consumers are being lured in for unobvious exploitation.
• snickell 1y ago
  What scares me is that the obvious pool of money to fund the deficit in the cost of operating of LLMs comes from the most subtle native advertising imaginable. Can you resist ads where, say, AirBnB pays OpenAI privately to “dope” the o3 hyperspace such that AirBnB is moved imperceptibly closer to tokens like value and authentic?? How much would AirBnB pay for the intelligence everyone gets all their info from having a subtle bias like this? Sliiightly more likely to assume folks will stay in airbnbs vs a hotel when they travel, sliiightly more likely to describe the world in these terms. How much would companies pay to directly, methodically and indetectably bias “everyone’s most frequent conversant” toward them?
  • john-h-k 1y ago
    > Can you resist ads where, say, AirBnB pays OpenAI privately to “dope” the o3 hyperspace such that AirBnB is moved imperceptibly closer to tokens like value and authentic?? This would be a very impressive technical feat
    • JimDabell 1y ago
      Anthropic demoed something similar with Golden Gate Claude a year ago: https://www.anthropic.com/news/golden-gate-claude https://www.anthropic.com/news/golden-gate-claude