Generative AI took the world by storm in November 2022, with the release of OpenAIs SERVICES ChatGPT. One hundred million people started using it, almost overnight. Sam Altmanthe CEO of OpenAI, the company that created ChatGPT, has become a household name. And at least half a dozen companies are racing OpenAI in an effort to build a better system. OpenAI itself aims to excel GPT-4its main model, introduced in March 2023with a successor, likely to be called GPT-5. Almost every company is struggling to find ways to adopt ChatGPT (or similar technology, developed by other companies) in their business.

There’s just one thing: Generative AI just doesn’t work well, and probably never will.

Basically, the generative AI engine is fill-in-the-blank, or what I like to call “autocomplete on steroids.” Such systems are good at predicting what is good or believable in a given context, but not at understanding at a deeper level what they are saying; an AI is constitutionally incapable of fact-checking its own work. This leads to many problems of “hallucination,” where the system asserts, without qualification, things that are not true, while inserting errors into the bones of everything from arithmetic to science. As they say in the military: “always wrong, no doubt.”

Systems that are constantly flawed and indecisive make for amazing demos, but often are bad products themselves. If 2023 was the year of AI hype, 2024 will be the year of AI disillusionment. Something I argued in August 2023, at first doubt, more often felt: generative AI can be a fool. The profit is not there—estimates suggest that the 2024 operating loss of OpenAI will be $ 5 billion—and the valuation of more than $ 80 billion is not consistent with the lack of income. Meanwhile, many customers seem to be disappointed with what they can do with ChatGPT, relative to the extremely high initial expectations that have become common.

Furthermore, essentially every major company seems to be working from the same recipe, making bigger and bigger language models, but all revolving around more or less the same area, models that are just as good. of GPT-4, but not a better one. What that means is that no individual company has “moat” (the ability of a business to defend its product over time), and what that means is that profits are shrinking. OpenAI has been forced to cut prices; Now Meta provides the same technology for free.

As I write this, OpenAI is demoing new products but not actually releasing them. Unless it comes out with some major development worthy of the GPT-5 name before the end of 2025 that is better than what their competitors can do, the bloom will be off the rose. The enthusiasm behind OpenAI will wane, and since it’s the poster child for the entire field, the whole thing may die out soon.