‘The AI industry is ignoring how much people hate it’ | Ed Zitron

The thing about this is that the AI industry is ignoring how much people hate it. Why are they surprised when you spend years going, "Eh, eh, poking people in the head, eh, we're going to take your job. Eh, we're so rich. We get whatever we want." Regular people suffer it. And who isn't the AI industry that's threatening to take their jobs? The dangerous rhetoric does need to stop. And it starts with being honest about what AI can do. And it starts with stopping this evacuous, ridiculous conversation about AI jobs that does not relate to anything that's happening right now.

Today on the tech report, I'm joined by writer of Where's Your Educt and the host of the Better Offline podcast, Ed Zitran. Thanks for coming back.

>> Thanks for having me. So, let's start with the attack on Sam Wolman and the dangerous rhetoric around AI you talk about in your free free newsletter this week. Obviously, I know this this show is pretty critical of Samman and and all the other tech bros and AI bros that are out there, but obviously violence is not going to help make things better and those supporting the attack, which I I saw in the comments, are not going to find any support here.

>> Yeah. Um, violence isn't the answer. Don't support these attacks. Categorically think they're deplorable. It's morally reprehensible. However, the result like these attacks did not come from as Sam Orman suggested critical articles of him. Didn't come from the Rod and Farah piece, didn't come from anyone. I don't I must be clear. I don't think he's in any way referring to me. I don't have that much of an ego. I think the thing about this is that the AI industry is ignoring how much people hate it. And it's not just, oh, I don't like that chat GPT is kind of bad at stuff. I don't think it's about that. I think it's that when you look at a regular person's life right now, harder to get credit than ever, harder to get a mortgage, harder to really accumulate wealth in any way, shape, or form. In America, at least, the cost of college, cost of health insurance is incredibly high. Everyone is suffering and everyone's having trouble. You look in the news, insane stuff's happening. We may or may not be at war with Iran. We have constant fuel shortages across Asia. Like, there are really scary things happening. There's all of this austerity and restraint in a regular person's life. You go and look at the AI industry. Oh, they've raised a bazillion dollars. They've ra $122 billion for chat GPT. Oh, they can have as many data centers as they want. Oh, all birds can become an AIG GPU company and get a 600% stock bump. And then when you look at what the AI people are saying, they're saying we are coming for your job. Get ready. 50 Dario Amade 50% of white collar jobs will be destroyed in the next however many years or months. It's always changing. Sam Orman I think I'm I can't remember if this is a comment about him or whether he actually said it, but he said something along the lines of uh enjoy your job while it lasts. We'll have super intelligence by the end of 2028. It's antagonizing people. The dangerous rhetoric here is the constant misleading like it's a lie. The constant suggestion that large language models are taking people's jobs. The constant suggestion that everything's going away, that creativity will be automated. That monthsl long campaign about Sora suggesting that people in the creative arts would have their jobs replaced. Yeah, the dangerous rhetoric is the constant threat against regular people who are struggling to get by. And so, yeah, I to be clear, the attacks are deplorable. We cannot do it's disgraceful. Physical violence is wrong. Violence is wrong. However, why are they surprised when you spend years going, "Eh, eh, poking people in the air. E, we're going to take your job. Eh, we're so rich. Eh, we get whatever we want. Need to raise $100 million. Boom. That's nothing. We'll just pull out the couch cushions. Nvidia selling more GPUs than ever. Regular people suffering. And who isn't the AI industry that's threatening to take their jobs? The dangerous rhetoric does need to stop. And it starts with being honest about what AI can do. And it starts with stopping this evacuous, ridiculous conversation about AI jobs that does not relate to anything that's happening right now.

Do you think something like this was I don't quite want to say inevitable, but kind of expected, especially if you add in what the the suspect's lawyer called an acute mental health crisis on on on top of the the the motives that you you've just outlined. So, just to be clear, one of the inspirations based on his Instagram, the the perpetrator of the Molotov cocktail incident, one of his things he was reading was Elia Eliza Yudkowski's um if somebody builds it, we all die. These overly verbose dunes that write hundreds of thousands of words. I know I'm criticizing someone for writing too much about how AI is going to be powerful and we need to stop it and oh, and oh, it's all so scary. platformed by the New York Times, by the way. So, yeah, mentally unstable people in a society that is increasingly unfair to most people are going to get triggered by this kind of thing. And on top of that, yeah, I think it is inevitable that something was going to happen because when you corner people and you piss them off and you anger them and you attack them and you you're horrible about their jobs, Sam Alman said a few months ago that some of the jobs that would be taken away weren't real work. What do you expect? I'm not saying that it was inevitable necessarily that someone was violent, but at the same time, why did like these people believe that they're untouchable? They believe that they're somehow separate and above all of us. So yeah, what a surprise when you piss people off enough, someone gets someone takes you seriously. When you threaten your people's jobs and to change society for your benefit, yeah, what a surprise. Somebody takes you seriously. somebody mentally unstable, somebody on the edge at a time when a lot more people are being pushed to the edge. Go and look at there. I saw a video the other day of Vinland, New Jersey, the data center there. It's constantly going. If I heard that all day long, I would feel kind of insane, too. And quite frankly, I like to be clear still thinks of violence is deplorable and all that. But the AI industry shouldn't be surprised that they get all of this attention, all of this money and constantly threaten people and people act in a way that suggests that they are being threatened. There's one thing that again a lot of the media as you point out in your newsletter point to the the New Yorker expose or some of the bad press that was happening. again to to to your newsletter, you talk about the sort of AI doomer doomerism hype kind of thing and and the if I was to pick a a trigger moment, which I I don't think in any way it was solely up to this, >> but the timing of Claude mythos and the the conversation around that and it's so terrifying. It's going to break everything. We all should be scared about it. It's broken out of a sandbox. It's coming to get you. That kind of language is exactly the kind of thing like you say that somebody who is already struggling and a and in in this kind of like situation where they're on the edge that you described that that to me feels like a a much more likely like a scaremongering switch flip than somebody insulting Samman or not even necessarily insulting Sam >> just repeating things that people have said was uh to quote Wall street too. Uh Gong Gecko says to someone, "Uh, you stop telling lies about me. I'll stop telling the truth about you."

So, here's the thing. Mythos is a great example. I'm glad you brought it up because it's exactly the problem. Mythos is didn't break out of a sandbox. What happened was it wasn't even on the same computer instance as the thing it was playing with. It was in a sec. The model was on one instance. The thing it broke out of, broke out of, which it was instructed explicitly to do. nothing to do with reality. This whole mythos is too scary. We couldn't possibly launch it. This whole look anthropic, they spread some of the most dangerous rhetoric of them all, which is this thing of, oh, our models are deceiving people. They're blackmailing people, which they are not doing. In the system card, it shows in every case the anthropist being like, it's the meme of tell me I'm tell you're alive. I'm alive. It's the same thing. And yeah, I don't know if I if someone is I was a mentally unstable person. If I was sitting there scratching at the walls, feeling like everything was going wrong and I kept hearing on the news that there and I read these books that said AI was going to destroy me and I saw on the news that AI was going to destroy me and now there's this model that's too scary, but they only get you'll notice they only gave it to the rich and powerful. Yeah, this might actually twist somebody up inside. I I this person shouldn't have done it, but I can see why it happened. And the AI industry is going to try and pretend that this is somehow not anything to do with what they're saying when it is. It is their dangerous rhetoric. And it is the mainstream media's support. I cannot be clear enough that the mainstream media also spread this dangerous rhetoric. They are the ones that have since 2023 spread the myths about AI. GPT4 back in 2023 they were saying oh it deceived the task rabbit it didn't literally they were copy pasting the prompts from a large language model into task rabbit and it's not even clear within the GP4 s um system card whether they actually even did it they kept mentioning theoreticals that's the thing the media could easily have pushed back on that instead they just published the the scary stories of oh AI is this AI AI is doing that it deceived this person it deceived that person oh we need to have the AI 2027 in the New York people in the New York Times. Daniel Kokij I think his name is the a goddamn philosophy student not even a scientist but because he wrote a very long blog AI 2027 we need to have him in there. The dangerous rhetoric comes from the media and it comes from the AI companies. The actual way to stop things like this happening is to have a fairer society. The other way is to start talking about this as normal technology. Large language models are normal technology. They are not magical. They are not even artificial intelligence. And as long as we keep letting the media spread this, as long as the media keeps falling for it again and again, they have they have thousands of fell for it again awards. As long as we let that happen, more people like this are going to get triggered. I hope more I don't think more of this. I really hope it doesn't. But we're at a time when the pressure on the 99% has never been higher and the threats have never been more dire. And we are sitting here worrying about whether Ronin Pharaoh inspired this. And that is an intentional deception with the intent of moving the blame away from the people that should be blamed, which is the AI companies and those that help spread their message.

Another side to this as well is that I mean there's the AI psychosis stuff and on top of that the AI fueling people to not just take their own lives but to commit acts of violence encouraging people to do these things. And there's I I just did a video on it. Um they they speak uh that there's a there's a bill in Illinois which is looking to limit the liability of AI companies in in c mass casualty events as long as they're below uh 100 which is just an arbitrary number obviously >> at a very high number as well. Yeah. um that it's just it feels a little hypocritical to be on the one one hand saying this person I mean it is it is unacceptable but then somehow AI convinced this person to go do a a mass shooting but because he only killed 40 people we're we're not we're not at fault for that there's no liability here for this >> so I I must be clear the son of Sam Alman situations the people being triggered in AI psychosis to kill people. I didn't see Sam Orman talking about that once. I didn't hear Sam Alortman being worried about the dangerous rhetoric spread by chat GPT itself. That's cuz he doesn't care. That's cuz he doesn't care. He doesn't That's not a problem to him. It It's bad when it happens to him. It's fine when it happens to other people. Actually, if the AI companies want to change something, they should immediately stop anthropomorphizing models. models should not respond like human beings or they should drastically limit that. They should pull back all of it. That crazy stuff we had when they destroyed GPT40 when they put put out out back like old yellow when people freaked out should have been a wakeup call that they should these things should respond simply like a terminal window but not with any of this like ohiculating splines or the cutesy language. You got to strip all that out because what it is twisting people up. They should not have conversations. You can and I know people who say, "Well, I need to brainstorm with it." Call a friend and if you don't have any, make one. It is harder to make friends than ever. I take I I'm clear about that. But you know what? Go on Reddit. It's better than chat GPT. If you want to talk to an actual person, but that's the thing. They don't want the AI companies do not want to talk about the real problems because then they'd have to come up with real solutions. But they don't have any. They don't actually care about any of this stuff. they are doing the bare minimum because the media doesn't push them. The media actually doesn't hold them accountable. I was shocked to see in a few stories saying things similar to me that the dangerous rhetoric came from the AI AI labs but they all pulled their punches. They all said yeah well all of the boomerism it's quite scary. No, it is scary but they have profited off of it. Sam Alman just to be clear that book I mentioned if they build it everyone dies whatever it's called. Eli Yudowski Samman said in 2023 that he might be worthy of the Nobel Peace Prize. That's that's their they are fully connected like these people are fully connected. Yodowski also in the Epstein files had a phone call with Jeffrey Epstein. Fun that. Um also Larry Summers used to be on the OpenAI board. Also good pal of Jeffrey Epstein. We had that. Putting all that aside for a second. It's just the media doesn't really want to get its teeth into this because the media knows it's partly responsible. The AI labs are doing the bare minimum because they know the media is not going to push them. If we had a functioning tech and business media that wanted to put pressure on them, the attack would be stop anthropomorphizing them. Stop doing this AGI thing. Stop pretending it's what you're working on because it's not. Large language models will not lead to AGI. Stop trying to build a conversational AI that has conversations. Limit it because they are dangerous. People say, "Well, I want to be able to do that. It's dangerous. It's hurting people. It's driving people insane. People are dying. I I don't know what else to say. Stop it." You can still have clawed code that responds conversationally. You can still have that um NLP. You can still have something that responds to regular text. That's fine. But these things need to respond and act like computers again because that's not what they're doing. They are intentionally deceiving people, twisting people up, and people are getting hurt as a result.

>> That addresses certainly the the AI psychosis side, there's also the this fear of being left behind that that ties in with the AI is going to take my job anxiety and while we're trying to be a bit more constructive, what what could people like Sam Alman, Darday do to reassure people that they wouldn't be left behind? And not just words, obviously. I think we're past words now. I think things need to be done a and shown that we're not just going to say, "Oh, you can have a 4 day work week sometime in the future. Don't worry about it."

>> I don't think they're willing to do what needs to be done because the actual answer is to section corporate profits. If they had them, which they don't, they don't have the profits. They can't put any aside. Uh they claim to do public benefit with Open AI actually create a public benefit. The stock is going to be worth assuming they went public question. Open question there. Uh be worth over a hundred billion dollars. I don't know. probably do something about hunger. I think what was it the quote that Elon Musk was given several years ago? Six billion dollars to end world hunger. Yeah, rip that off Sammy. Come on, >> Samuel Glamuel. Let's do it. Fix world hunger. Go. Um I also think that all of them should roll back their language direct. I think there is a linguistic change. I think they need to say that we regret that we've ever talked about job loss, that we should stop having it. I know from personal experience that Dario Amade is very very very proud that someone meaning him is willing to have the AI conversation the AI jobs conversation enough rip out all of them marketing hype stop it it's not it's not worth it anymore but they will keep doing that if they really truly actually wanted to do something they would work if these are going to be the biggest most wealthy companies of all time they would be building in I don't know Medicare all for America. They like just do something. Just do something as a social good, build housing. Stop buying GPUs, build housing. Like there's you actually want to make the world better. If you actually believe that there needs to be UBI, then your company needs to provide it. I don't like that idea. Government should do that. We should not rely on capitalism to fuel our our economies in general. But if they want to do something, they could do something instead that they they do these vacuous. Oh, we're doing $50 million grants that no one can seem to access. We're giving away millions of tokens. We're doing these things that sound good to the media that the media can put in an article headline and then never think about again. Oh, it works out great for everyone other than everyone else. And so they won't do this, but they should turn around and just say, "We are not going to talk about jobs anymore. That's great stuff. We're not going to talk about it. This is not our place. We're going to talk about our products. We're going to talk about them in regular ways. We'll talk about we're excited about them because of this." But instead, what they're doing is this endless cycle of, "Oh, the models are scary. We couldn't possibly release them other than releasing them to banks. Banks and cyber security. We're going to give them to the most trustworthy people in the world, banks. We're going to get make sure that JP Morgan can use Claude Mythos. Also, here's the thing. Got 40 different companies using Claude Mythos now, if not more. Why haven't we heard very much about it? We had one of these AI safety research groups say something about it. But I don't know. I wouldn't I wouldn't trust Dick Dastardly about the amount of fleas on Mutly. So, I'm not going to trust these people to tell me whether something's good. It's always better in an indeterminate way as well. But this is meant to be a big scary cyber security thingy. It's scary. It's too powerful. We couldn't launch it. But now what? What's the What am I meant to be freaking out about? I luckily have therap I have therapy. I have great mental health. I can look at this and be like, "Oh, this is a marketing ploy." Somebody who is sitting there scared, who has been mainlining AI doomerism, people that read less wrong, people that are freaking out about this, people with real mental health issues. Yeah, I would see the mythos thing and I'd freak the hell out. Not because it's tangibly scary in any way that I can measure, but because it confirms the biases that I have that have been laid for years by these AI companies that my job and my health and my future are at risk.

>> Unless there's anything more you want to talk about on this. Should we?

>> Let's move on.

>> One headline that caught my eye was OpenAI pulling out of another data center deal, this time in Norway, which I think makes it the third time this has happened that the Open AI has pulled out of a data center deal about in as many months.

>> Yep.

>> How how would you describe the the state of the the data center buildout right now?

>> Not good. So, I think when all of this is over, I intend to do my detailed notes piece where I write down where everyone was wrong. But the big thing that I want to say is why is the media not saying anything about the fact they all got hoodwinkedked about the $500 billion Stargate data center project? Donald Trump, Masoshi's son, Larry Ellison, Sam Orman, Glammy Sam Orman up in the White House claiming that the Stargate data center project was going to do all this stuff. He hasn't really done much of anything. We've got a few steel beams in Wisconsin or Michigan. We've got maybe some ground being dug into in Shackleford, Texas. And then we have a dead project in Norway. Dead Project in London. The extension isn't happening and maybe Microsoft's taking it. Also, Microsoft is taking over the Norway one. Very weird. All very strange. The next on the Death Watch is Stargate Argentina, which was always a letter of intent, by the way. So, dead. Like just that. I'm just going to assume that one will be announced as dead if they even ever bother. And I really do think like this is it's obvious that they were just making stuff up because the Stargate Norway is fascinating. They never had an agreement. They did an announcement saying today we are launching Stargate Norway. No, you're not. You lied. You could say, you could have said announcing then you wouldn't be lying. But you said launched. No media coverage of that. No media coverage of the fact that they've just been lying. No media coverage that everything's behind behind schedule. just a lot of it's almost as if the business and tech media is able to consider facts in a vacuum but never together. The answer is is that data center development is hard and also open AI doesn't have resources like that. They don't have the they can't raise credit. They have billions in losses like no one's going to extend them credit. They don't actually have much infrastructural experience because building large data centers over 50 megawatts is a relatively new maybe three or four year uh old field. And on top of that, they let this take forever. Where at the time when this was announced, I w I genuinely went to people and being like, "Hey, where are they building this? Do you have a place? How are they building this? Who's funding this? What's happening here?" And no one could answer. And no one bothered to ask and no one bothered to say. I think that it's disgraceful and a black eye on the media that this happened and no one is saying anything because I think readers and listeners are very well aware that this hoodwinking keeps happening that Charlie Brown keeps missing the football. Little American reference there. Um, and it's frustrates me. It frustrates me because for years I've been covering this and every time I've said something like, "Hey, Stargate Norway doesn't seem real." People will genuinely attack and say, "Oh, yeah. Well, yo, yo, you don't believe in progress? Uh, OpenAI would just like Yes, they would. They They're doing it right now. The question to ask now is, how much can we trust anything that Open AI says anymore? They talked about that all the good stuff they're going to do with the nonprofit. Can we trust that? Can we trust that OpenAI is even going to release that consumer device that they spent $6 billion in stock on? Do you think that they're actually going to build a super app? They were talking about for months they're building a super app that combines their browser, codeex, and chat GPT. Where is that? Anyone seen it? Because I swear to God, if it turns out that they announced that super app and it was in the Wall Street Journal and it was in the Times, it was all over the place, if I have to if I have to see that not happen, I'm taking some SC. I'm I'm so tired of the open AI marketing reality distortion field because this company is they are inherently dishonest. Sam Orman is an inherently dishonest person. Not just based on the New Yorker, but based on him announcing Stargate at all. It does not exist. It was never incorporated. This is a company that lies. And now Anthropic is doing very similar things. They signed a deal with Broadcom. They signed a deal with Corey. I think anthropic is just open AI 2 and I think open AI is we work two which makes anthropic I don't know maybe we work three we'll have to get back to that one.

>> why do you think Samman or open AI is having to draw back its data centers data center spending? I mean, it's obviously just had a big funding round which is somewhere between 60 and 122 billion not quite sure yet. Is or is it because they've over estimated demand or they're just not quite sure they're going to have the money going forwards?

>> So I think that they just said these data center projects without really checking whether they were real. I think that they keep say like how many different like that hundred billion dollar round with Nvidia that never happened. I think that they are helping companies do stock manipulation. I think that they do announcements to help bump stock. I think they do it to help companies raise money. Ncale was able to raise money off of the London deal. Nothing happened. I'm sure Achre aka ER who's the Norway partner. I'm sure that they've raised money again based on the theoretical idea of Stargate Norway. I don't know. Do we have financial regulators who are awake at the moment? I think that none of that was ever happening and it was all announced with the understanding that Mr. Alman would get someone else to do it. And if they didn't do it, well, they'd come up with some excuse such as, "Oh, regulation. Oh, oh, there's too much regulation in England. We couldn't possibly. The energy costs are too high. Wow, Sam, the power's expensive in England. None of us knew that. If only you could have, I don't know, talked to anyone from England. Perhaps they would have let you know. Oh, you mean there's regulatory problems? No, there's not. They announced the core wave and the the Stargate London stuff with Open AI. I'm sure Kar well Kstar asking K star for an answer on something might be difficult. But nevertheless, the point is all of these things were theoretical to help build hype for open AI. As they fell apart, they knew that the media would not call them out on it. They knew that investors would not call them out on it. These are useful tools and deceit and acts of deceit that OpenAI uses to pump its stock and others. And quite frankly, I find the whole thing disgraceful. And I think that I have to wonder how it's even possible that OpenAI makes even a quarter of the things they've said happen. And the answer is they probably won't. I wouldn't be surprised if one of these big data center projects in Shakford or Wisconsin gets kind of railroaded and ends up stopping. I wouldn't be surprised if they have to cancel some capacity from Corewave. Maybe their Serapus capacity. The 750 megawatts of compute from Caprris. Don't know where those don't know where those cerapus chips are. They haven't said 750 megawatts is a lot of power. Where are they? I'm sure that they'll just hope nobody asks about that again. And they won't. The people that actually interview Sam Alman won't. And this is how a con works. A con works by working and exploiting the most powerful, most easily misled people and also the biggest cowards. Cowards are deeply important to the OpenA. the cowards who cannot say that Sam Orman is a liar, that cannot say that OpenAI is misleading people. They know the media won't do it, so they'll just do what they want, and no one's going to stop them. We don't have financial regulators with any teeth anymore.

>> Let's let's go back to your your free newsletter from this week. you go in to talk about some of the the service reduction complaints that people have been having about uh Anthropic in particular, the speculation around they've been calling it bugs, user user error, settings problems, that kind of thing. I mean, maybe I'm getting this wrong. I mean, what what do we know it is from from not the people selling us the thing? And I is it not kind of go beyond in shitification to renegotiating already done deals if they are limiting rates and yeah giving people what they kind of didn't tell them.

>> Yeah. So just to be clear the large language model companies have been misleading their users for years. They should have never sold a monthly subscription. They cannot afford it. It is unsustainable. the massively subsidized tokens, meaning that when you pay 200 bucks a month, you can get two and a half thousand dollars worth of compute. Just to be clear, unclear if that's the case anymore. So, as Anthropic hits capacity issues, they are having to they claim that they're not making the models dumber. Claude Opus 4.6 is apparently materially worse and burning way more tokens. And literally, their response is like ah is fine. Uh the their service has been down three or four times this week. And the people will say, "Oh yeah, well this is just demand. It's demand. Anthropic has so much demand." Here's the thing. Anthropic isn't turning away customers. Anthropic is not shutting off their payment portal. They're still signing people up. Which means that this is not a demand issue. This is not about Anthropic not having enough capacity to make money. By which I mean revenue, not profit. No, this is Anthropic is greedy. Anthropic does not have enough capacity to satisfy their current customers. And they keep adding more customers so they can go public without having any plan other than we ordered some more compute in the future. And the thing is people are saying, "Oh, this just proves that they're super popular." Oh, it doesn't. I don't believe they're $30 billion of annualized revenue for a second. I don't think it's possible. And we'll find out eventually when they file their S1. But what we do know is that Anthropic is making their product worse. They are reducing those rate limits. So you have you can use it way less in in the daytime. I think it's between 5:00 am and 11 a.m. Pacific. I'd have to look. And the thing is this is only going to get worse. Enterprise customers on Claude now have to pay the actual token rates. Uber's head of AI said that they ble blew through their entire yearly AI budget in three months. This is now getting to a point where the actual costs are being reflected in the services and these companies are not run in any kind of responsible way. The idea that if you bought let's say you were using at the end of last year there was 2x increase just for Christmas. You bought an annual clauses subscription on December 27th 2025. The product you use there does not exist anymore. The rate limits are more than haveved. You have significant daily rate limits that didn't exist. The models are dumber and think and the service is down half the time. I mean, I don't know how this isn't considered fraud. I'm not saying it's fraud. I'm saying I don't know how there is not a competitive body looking into this because the service that you bought even 3 months ago is materially different. The functionality is different. Think about it like this. If you built workflows based on being able to burn a certain amount of tokens a couple months ago, you cannot run those same workflows within these rate limits. The product does not function in the same way. What are you meant to do about that? How is there no competition committee looking into this? And there isn't because that would require functioning regulators. Things are only going to get worse from here. I would not be surprised within the next six months if anthropic just says any teams and enterprise subscription has to pay token rates. I wouldn't be surprised if in the next year they actually just force everyone to pay the token rates. And I think at that point we know that this whole thing's wrapping up because I the biggest deception from Chad GPT and Claude is that for 20 bucks a month you can have a functioning AI service when in reality those 20 bucks people are spending 80 to 100 bucks a month worth of tokens. There was never a point in which a subscription-based AI service would make economic sense. And we are now seeing what happens when they try and make those economics work. You get a broken product that sometimes kind of works that does not work in the same way it worked 3 months ago and probably won't work in the same way in 3 months from now, even with any kind of Opus 4.7 release, which just came out as we were talking. It's it's a disgrace. And again, where's the media on this? The media shouldn't be right. They shouldn't be saying, "Oh, yeah, people are upset. People, they should be saying people have been misled because they have."

>> What happens when, assuming you're right, and and Anthropica, the Open AI, and these companies start charging token rates for for for what they're selling. May maybe not to consumers at least, but maybe to enterprise. I mean, it doesn't seem like it's affordable to me. I I maybe I don't understand business.

>> No, no. It's it's fair to ask that because at this point, let's say you're a huge company with thousands of engineers. Are you really going to spend tens of millions of dollars a year on something that doesn't really seem to make them better in a way that makes you money or makes the product better? You can kind of see this happening, by the way, with with Claude themselves with Anthropic themselves. Even Claude's desktop app, go and look at Theo Brown on YouTube if you want to see this. The Claude desktop app is broken. It sucks. It's genuinely bad. This is what it looks like when you build everything with AI. The tech stack across the entirety of the tech industry is getting worse. It's getting worse because of these tools. These tools that also are costing what, 10% of headcount, I read somewhere, if not more, possibly rising to as much as 50%. What are we doing? Why are we doing it? So that we can have the most annoying people in the world go, I write everything with LLMs now. I'm I don't write any code at all. Wow. Is the code good? You don't know. You're not reading it. You just hit the button, ship the code. And that's the thing. You are now, I think, in the next few months especially, you're going to see these things where companies start having to really reconcile with the with the actual value of AI because before this point, they could get away with using the subsidies, the subsidized subscriptions. Now that they can't do that, I have to wonder if they're going to give much of a damn.

>> Well, Ed Zitron, thanks for taking the time.

>> Thanks for having me. If you enjoyed today's episode and you want to hear more of the tech report, please consider liking and subscribing.