What if artificial intelligence never becomes much better than it is today? This is a question recently posed by Cal Newport in an article about the current development of AI for The New York Times. Newport is, among other things, a professor of computer science at Georgetown University and a well-known author of technology books, the most famous of which are ‘Deep Work’ and ‘Digital Minimalism’.
In his article, he explains the technical limitations of artificial intelligence models and the reasons why, in recent weeks, investors have been withdrawing funds from AI companies, and many companies fear they may never recover their investments in AI.
Is a New Dot-Com Bubble Forming?
The difference between each previous version of ChatGPT (1-4) was enormous, but the leap between ChatGPT 4 and ChatGPT 5, officially presented on August 7, 2025, did not meet expectations. As Newport explains, the ‘fifth’ primarily relies on pattern matching rather than true reasoning, despite earlier hints that this could happen with AI in the near future. And lo and behold, reasoning is still required for humans in everyday work, which remains exclusively in the domain of humans.
As a result, people are generally dissatisfied with the responses provided by ChatGPT, whether it concerns text, numbers, or code, and investors are also unhappy, as they have largely been withdrawing investments from AI companies in recent weeks.
Even the world’s most valuable company, Nvidia, whose market value exceeds four trillion dollars and accounts for eight percent of the S&P 500 index, recorded a 3.8 percent drop in stock prices last week, followed by an additional 2 percent this week, after disappointing revenue forecasts, reports the Financial Times. The Guardian previously reported on the decline in stock prices of Palantir, Arm, Oracle, and AMD.
Market events also affect the perception of investments in AI and the expected returns. Some analysts refer to this situation as an ‘AI bubble’ that could burst, comparing it to the dot-com bubble of 2000.
Investors First in Line
The existence of the bubble and its significant inflation is confirmed by Davor Aničić, director of VelebitAI and a member of the Board of CroAI, the Croatian Association for Artificial Intelligence.
– The AI bubble is very different from all other bubbles so far. It is most inflated in a narrow circle of the largest companies betting that by investing billions, they will create general intelligence or even superintelligence, which will give them a lasting and defendable advantage over all other companies and significantly compete with all humans in intellectual work. I believe that this part of the bubble is about to burst because, in the last year, there have not been significant advancements with the current architectures of large language models – says Aničić.
If the AI bubble does indeed burst, Aničić does not believe it will significantly affect the broader business community outside of Silicon Valley and the largest AI players.
On the other hand, today, 25 years later, we are indeed smarter, so the vice president of CroAI, Ive Botunac, believes that a burst is unlikely.
– If it ever happens, I believe that the first to be affected will be investors, whether they are venture capital funds or ordinary people like us who invest in the stocks of such companies. And if we look at the past and the early 2000s, we see that the companies themselves were affected, and consequently, the users – says Botunac, who does not believe that AI will ever be able to replace humans in the way the general public perceives.
