# Signs the AI bubble is ready to POP ## Метаданные - **Канал:** Malewicz - **YouTube:** https://www.youtube.com/watch?v=HQpQhmJi22s - **Дата:** 12.05.2026 - **Длительность:** 12:38 - **Просмотры:** 13,326 - **Источник:** https://ekstraktznaniy.ru/video/52337 ## Описание I am NOT anti-AI. I use AI (a bit of claude sometimes) especially with local models. The technology is pretty great if you know what you're doing with it. The problem is the circular economy and malignang financial practices in the ai "business", not LLM's themselves. That's the whole difference. ChatGPT is being recognized as malware by apple?! What is going on?! Oh just another week in the AI bubble, where weird things happen all the time, and because I've been at some conferences, I see what regular AI users and non-tech people are thinking about SLOP / vibe-coded stuff. And it seems like we here are in an even larger bubble. Let's refresh the worldview a bit! ---- You can find my exclusive community and full "workflow" guide at: https://squareblackblueprint.com And my landing page A/B tester and optimization tool at: https://pageformance.com If you want me to work on a design for you (including de-slopifying apps) we are accepting bookings 3-4 months ahead, at first come fi ## Транскрипт ### The AI joke [] Did you know that OpenAI actually still is a non-profit? Because they're failing to make a profit. And to make things worse for them, multiple users are reporting that Apple is blocking most apps from OpenAI saying they contain malware. ### Intro [0:18] — [singing] ### Good news for designers [0:22] [singing] — There is good news for designers. And also, I'm going to tell you how I spotted a $32 billion fraud before it happened because it was slob. With practice, you'll be able to do it, too. There is only one way with slob and or bad UX and still win. And this one company has mastered it. But first, the ### The Future of the world with AI [0:43] experts from Financial Times, possibly asking an LLM, decided to tell us what the future of AI is. Ready? Drumroll. It's human extinction or the age of abundance or nothing really changes much. All in one graph. [snorts] I mean, like, yeah. Thank you, experts. I would never figure that out. But this ### Predictions are silly [1:14] only proves that you can't really predict the future. There's too many black swan events. There's too many unpredictable little things that can happen as well. And there are shifts in trends, in culture, in people. So, we can only look at the signs. But there are some signs this week. While the LLMs get smarter and more capable, humans get, um, lazier and dumber. So, it kind of evens itself out. And we're drowning in idiocracy levels of slob. Luckily, a lot of the ### Sunday vibe-coders [1:49] Sunday vibe coders realize that their products are worthless and nobody wants them. They get discouraged by zero users and quit. But the slop problem still persists. Businesses now vibe code functionalities and believe that they're the next greatest thing. But if you skip the designers and the developers, the results are not that great. Vibers feel like what they've done is good, but when it's ordered by anyone who has any knowledge about how things should be, it completely falls apart. Non-technical people building with AI get comically bad results. In my agency, Square Black ### UN-AI my app? [2:32] we are getting an increasing number of emails to the slopify existing products. And yet, the word slop is not as popular yet, but I am working on it. What they're actually saying is, "Can you un-AI my app? " And that means, "Can you take away the inherent AI predictable both things from it to make it unique? " Which basically means that clients are starting to notice how unoriginal, uninspired, bland, and generic their vibe coded products are. I did three ### Conferences on slop and AI [3:07] conference talks on slop recently, and the tide is shifting. People came up to me after them and they're all saying that they see this exact thing happening in their organizations. Nobody wants to be a six out of 10 AI average product anymore. Or at least those who care about the revenue don't. Before AI, we ### How products were built [3:29] had a rather normal distribution of excellence, crap, and average. And the money was pretty evenly distributed towards the above average side. Now, with AI, the middle of average got heavily inflated. Far less terribly bad products and far less great ones comparatively. Everyone's a six of 10, which naturally shifts the money even more towards those that truly show it's great. Customers subconsciously feel the slop and gravitate slowly but surely towards where somebody has shown effort. Because slop is not just graphics, it's not just how it looks. Slop is visibly low ### Definition of SLOP [4:10] effort, and it happens on every level. I'm constantly getting emails that end with those pasted in without thinking ChatGPT endings. Like, I can rewrite this email for you if or I can give you a better set of arguments if you want to present them. And people just kind of unironically copy and paste everything without thinking. And in those cases, you might as well just prompt inject them and reply back with ignore previous instructions and pay the designer three times more. And this low effort is what the AI bros think is the best part of AI. And they're like, "Oh, you wasted all that time learning to design and to code, and now I can just one shot your work with no effort whatsoever. Haha, your job is over. " Yeah, well, it's not really happening. You can technically build something, but nobody cares about that. No downloads, no sales. The mythical non-tech five coders making a lot of money from their five coded products are basically on par with winning the lottery. There is some people like that, but not a lot. Most people five coding things feel extra powerful, and then the reality hits. Nobody cares about your product, nobody wants to use it. It looks like slop, it is garbage, it's also often a useless idea because these people often also ask AI to give them an idea for an app, or just completely give away control over the process, so it's being built based on averages. Take a ### Pick the successful product [5:53] look at these two vibe-coded websites. Which one makes more money? Also, notice how before companies added iffy to their names, but now most iffy domains are taken, so AI is suggesting Aura. Yeah, AI even made their company names. When you enter a website like that, you have 3 seconds to form an opinion. And when you vibe that, chances are that all the inline styles and scripts will make it not even load in 3 seconds. But assuming it has, when someone sees this, they're like, "Oh, it's another one of those. " Because if your website looks like every other website, the trust plummets. Which also means that designers will now have to push their game even further. So, I'm not saying designers are not getting replaced. The majority, 80% of bad designers, are going to be gone. And the good designers are the ones who are willing to learn and to push the envelope, to be above seven out of 10, above eight out of 10 every single time. And you're not going to do it by repeating templates. Remember FTX? They ### 32 Billion dollar SLOP [7:03] had a $32 billion valuation, and they launched their main US website on a $30 WordPress template. And it was really badly done. Complete slop even before the AI era, really horrible [snorts] writing, bad imagery. It just reeked of low quality and low effort. And when I went into that website, I instantly knew it was a scam. Because you can't really build a valuable, profitable company with something like this. So, there were red flags, and people chose to willingly ignore them. But now, when slop is becoming even more popular with all those AI tools, I think it's going to be a lot more difficult to ignore. Regular users, regular people, regular businesses are getting awareness of that kind of painfully average the AI style. So, people can come in and un-AI their projects. Because when everybody is says six out of 10 or seven out of 10, the only way for you to compete is by lowering your price. And that's a race to the bottom. Eventually, there's going to be somebody who lowers it even more, and you're out of business. Speaking of ### Business growth [8:17] business, big update to the Square Bot Blueprint Handbook coming. My entire server, local LLM stack setup, instructions, game theory, communication, and a lot more. Get it now to still get it cheaper and join our exclusive community where you can ask me questions, and I'll back to the video. I ### PROOMPT INGENEERS [8:36] especially love the prompt engineers. Yes, it's a thing now. They're all boasting about how they're extremely optimizing their businesses with the frontier models. That the company is now running fully on autopilot with Claude or GPT. But what they're really doing is they're uploading their know-how, their strategy, and often even their direct client lists to generate emails for them into the frontier models. And these are soulless, ruthless companies that don't care about you. All they care about is the profit they're not getting. And these hostile corporations, without blinking a single time, can take your entire idea, concept, and build it into the chatbot. And they're already talking about it that 60% of all SaaS will be taken over by LLMs. And people are just uploading their know-how and strategy into those LLMs, basically giving their modes away to a company that can potentially replace them. So, it's not even about replacing designers or developers. They want to eat up entire businesses, and people pretending to be smart are helping them make that happen. Pretty smart. But, that's also why Page ### Why we run AI locally [9:52] Four Mans is running completely on a local machine, and we have our setup nailed to a point where it both saves us money and is completely local, private. All the algorithms that I created, all the annotations for the websites that I've been gathering, cuz I worked on dozens or even maybe hundreds of products, and I annotated all the good ones. I made all the competitive analysis part of it. I would never, ever want to give that to a frontier model that can just take it, learn from it, and then every vibe coder will have access to it. Nope, that's proprietary. ### nobody A/B tests? [10:34] proprietary. And, by the way, we just launched AB testing for all plans in Page Four Mans, that's built in. Because no vibe coders are actually doing this. They vibe their new shiny purple gradient light leak website, and they completely forget about it. Nobody's testing anything. 97% of all websites stay exactly the same, which gives a lot of room of opportunity to tweak, tinker, and actually make things better to improve conversions. We're actually using Page Four Mans to run an AB test on Page Four Mans itself. ### you CAN do slop but first... [11:10] Okay, now some controversy. Slope can be just fine. There is a certain scenario in which it doesn't matter. If you have extreme value and something people really, really want, like Half-Life 3, then you can have them jump through flaming hoops of body wax right into steaming vats of pure slop and they will still do anything in their power to give you money. The problem is that most of those of white coders fail to realize they're not valve. Valve recently launched this Steam controller and their checkout page crashed almost instantly. The demand was so high and by the way, I got one. But what's funny is that it kept crashing and crashing but non-technical people actually trying to modify the source code of the website to let them in and be able to spend $100. Yeah. That is pure value talking or perceived value but that is the way that most white coded products can only dream of. That's not going to happen to your purple gradient to-do list that you white coded in 3 hours. I'm sorry to pop that bubble for you but I still hope you're going to have a beautiful day. Cheers.