tldr; bubble bursts Q2, economic ruin, death, despondency.
An article dated three years ago in Business Insider makes for interesting reading in the context of the hype around AI.
In it, Jyoti Mann made the observation that Meta had, at the time of writing, invested some $36 billion in something called the Metaverse (a word capitalised back then; It Was Important), with the company under Mark Zuckerberg’s leadership proclaiming that the virtual online world was, without doubt, going to be the next big thing in technology.
True, Zuckerberg’s enthusiasm for the project was shared by few people apart from him, especially as it turned out, his own shareholders. Even in 2022, the money-people were advising that the company focus on profitable projects rather than something that was likely a bunch of specious claims made by a tech ‘visionary.’
The main thrust of the Business Insider article was that some of the most world-changing developments in technology have cost companies a fraction of what Meta was ploughing into the Metaverse. The iPhone, for example, is reported as having cost Apple, at most, around $3.4 billion in R&D, marketing and manufacture. Android cost Google $50 million, plus another near $7bn on research and development to bring a workable mobile operating system to market to compete with the iPhone.
As of April 2025, Meta had plans to invest an eerily-familiar $35-$40 billion into AI, and a great deal more since. According to the Financial Times [paywall], the company’s shareholders, like those at the majority of big technology companies, were hugely concerned back in April 2025 that the predicted profits from AI investments were, and were likely to remain, elusive.
There are concerns among ordinary people that their data, even that posted on ephemeral social media platforms, will be used to train AI models. That worry is writ large in the fury of professional ‘content creators,’ AKA artists, writers, film-makers and musicians – people whose copyrighted creations are what makes life worth living. AI can only produce literally average, barely-worthwhile facsimiles of existing data, and the more large models create and publish, the worse the learning corpus used to train the next generation of the technology becomes.
The advantage of AI is that it can ‘understand’ human language and respond very coherently, as long as the accuracy of its responses are not examined too closely. It’s like listening to a thick bastard who was educated at Eton. Sounds right, talks shite. Even in areas that initially proved promising, like summarising large inputs (document summaries, web search), the results are best-guess, probability-salted-with-randomness, rather than derived from insight, incisiveness, or ‘intelligence.’
In an effort to prove that investment in AI is founded on anything other than hype, early attempts at monetisation are doing nothing to convince people that AI has value. A Google or Bing search now yields results that are little but advertising materials, and now, ‘AI-powered’ results come first, whether you wanted them or not. Microsoft rolled-back an AI desktop ‘feature’ that helpfully recorded users entering online banking passwords, and is starting to redeploy a ‘safer’ version. Fingers crossed, then.
The tragedy is that the narrow use-cases in which AI is of proven value, such as the flagging of anomalous medical scan results, gene therapy, finding out how the universe works, and so on ad infinitum are not the areas that are the targets for big money investment. This year, when shareholders finally get tired of investing in flawed technology that literally produces nothing new, we will see the end of AI hype. Artificial intelligence is the 2020s version of flying cars or personal jet-packs, NFTs, or the now lower-case metaverse.
Meanwhile, there will emerge a use for AI technology that actually does something, like works out how to stop the icecaps melting, or how to prevent particular gene-carried diseases from being passed down through generations. That may attract a little investment cash, and the world may be a better place. But summarising emails ain’t all that – has everyone forgotten how to skim read? Funny animal videos? Anodyne marketing messages promoting stuff on social media?
Apple’s $3.4 billion iPhone was a distillation of existing technology, refined after many hours of human ingenuity and creativity. AI is a trillion dollar distillation of existing data that’s randomised just enough to sound a bit original, and that simply has very little practical use. It’s like someone inventing a box with flashing lights on it, then trying to convince everyone that what they need, so as not to get left behind, is a box with flashing lights on it.
Like many technology inventions, it’s a cool thing that everyone needs, right? Not sure why everyone needs it, but they just do.
The smartphone changed the way the world works, for better or worse. AI won’t change the world other than to create a lot more hot air, which sounds harmless enough until you realise that much of the hot air is the result of the effects of CO2 in the atmosphere.
Prediction #1. The AI bubble will burst around May-July 2026.
Prediction #2. AI will quickly be rolled back out of its current front-and-centre billing in every f*cking piece of software.
Prediction #3. Any users of AI, post-bubble-burst, will be paying a lot for it soon. Initially, 4-6 times more. Give it a couple of years, and the average user will be asked to pay $1,200-$2,000 per month for what they get, at the time of writing, for $200 a month for ChatGPT. Oh, and watch Office 365 jump in cost from around $10 per user, per month, to around $60-$120 a month for a ‘Copilot Enhanced’ version of the bloated suite.
Prediction #4. Apple, of all the big technology companies, will do the best out of the whole debacle, having not invested significantly, or at least, to the extent that Google, Microsoft, Oracle et al. have done.
Prediction #5. Professional roles that are a bit crap will be replaced by a technology that’s a bit crap. Writing social media posts, fr’example. Writing product descriptions for 10,000 SKUs. The kind of thing that you’d rather tear your own fingernails out than have to do.
Prediction #6. Satya Nadella, at the time of writing shifted sideways to become the head of AI cobblers at Microsoft, will take the fall for Microsoft as things go tits-up. He will be released from the company in order to spend more quality time with his personal wealth.
Prediction #7. OpenAI will be subsumed by Microsoft.
Prediction #8. Massive stock market crashes around the world in Q2-Q3 2026. Belt-tightening all round for the vast majority of the world’s population under the guise of austerity.