Generative AI takes the world by storm with release in November 2022 OpenAI Service chatgptPractically overnight, ten crore people started using it. Sam AltmanThe CEO of OpenAI, the company that created ChatGPIT, became a household name. And at least a half-dozen companies have raced to OpenAI in an effort to build a better system. OpenAI tried to outdo itself GPT-4Its main model, Introduced in March 2023With a successor, likely to be called GPT-5. Virtually every company hurried to find ways to adopt ChatGPT (or a similar technology created by other companies) into their business.
There's just one thing: Generative AI doesn't really work that well, and it probably never will.
Fundamentally, the engine of generative AI is fill in the blanks, or what I like to call “autocomplete on steroids”. Such systems are very good at predicting what might sound good or admirable in a given context, but not at understanding what they are saying on a deep level; AI is constitutionally incapable of fact-checking its own work. This has led to huge problems of “hallucinations”, in which the system makes claims without qualification that are not true, while introducing blatant errors into everything from arithmetic to science. As they say in the military: “Often wrong, never doubt.”
Systems that are often flawed and no doubt perform brilliantly, but are often poor products in their own right. If 2023 was the year of AI hype, 2024 has been the year of AI disillusionment. The argument I made in August 2023, fueled by initial skepticism, has been realized more often than not: Generative AI may prove useless. There is no profit there-suggest estimates OpenAI's 2024 operating losses could be $5 billion – and a valuation of more than $80 billion is not commensurate with the lack of profits. Meanwhile, relative to the exceptionally high initial expectations that had become common, many customers seem disappointed with what they can actually do with ChatGPT.
Furthermore, essentially every major company appears to be working by the same method, creating larger and larger language models, but all ending up in more or less the same place, with models that are as good as GPT-4. But no. Much better. This means that no individual company has the “moat” (the ability of a business to protect its product over time), and this in turn means that profits are declining. OpenAI has already been forced to cut prices; Now Meta is offering similar technology for free.
As I write this, OpenAI is demoing new products but not actually releasing them. Unless it comes out with some major advance worthy of the GPT-5 name before the end of 2025 that is decisively better than their competitors, the roses will remain bleak. The enthusiasm fueling OpenAI will wane, and since it is the poster child for the entire field, the whole thing may soon fall apart.