Panic connected us: the world was talking about the collapse of the "AI bubble"

The advent of artificial intelligence (AI) in recent years has greatly changed people's lives both at home and at work. He helps many people with daily tasks, while for some, investments in AI companies have become a source of additional income. However, recently, investors and analysts around the world have begun to talk about the collapse of the "AI bubble" and the problems of companies engaged in this field. Details can be found in the Izvestia article.

Prospects and the "AI bubble": what is it, when to expect problems

Recently, there has been a lot of news about the so-called "AI bubble". Even Nobel laureate Jeffrey Hinton, who has been called the "godfather of AI," has questioned the economic viability of artificial intelligence. In an interview with Bloomberg, he voiced his doubt that investments in this area will ever pay off.

According to him, in order to achieve financial success, companies would have to completely replace human labor, and this is impossible.

It is important to note that the issue of the crisis in the neural network market was raised back in October. Then Amazon founder Jeff Bezos and representatives of the International Monetary Fund (IMF) warned about it.

A report from the Massachusetts Institute of Technology is also indicative, according to which 95% of organizations have not received any return on their investments in AI initiatives. Researchers from Harvard and Stanford hypothesized that this is due to the use of neural networks by employees to create "work garbage", that is, content that disguises itself as good work, but does not have the essence necessary to advance towards the task. They also found that the prevalence of "work junk" can cost large organizations millions of dollars a year due to reduced productivity.

Last week, a number of media outlets started talking about the fall in AI companies' shares, as well as investors' concern that companies' excessive spending on neural network equipment may outpace real demand.

Modest incomes in this area are also recorded. According to analysts at The Economist, the total annual income of Western neural network leaders reaches only $50 billion, which is less than 2% of the $2.9 trillion that financial conglomerate Morgan Stanley predicts will be invested in new data centers from 2025 to 2028.

In addition, according to a report by the international consulting company Bain & Co., by 2030, AI businesses will need a combined annual revenue of $2 trillion to finance the computing power needed to meet projected demand. However, according to the forecast, revenue will be $800 billion lower.

The inflating "bubble" is also indicated by the growth of investments in AI startups, the number of which continues to grow. Even US President Donald Trump signed a decree on November 24 to launch the Genesis program, which is expected to accelerate scientific discoveries by using AI to analyze large amounts of scientific data.

The IMF has already compared the situation with the "dotcom bubble" of the 1990s and 2000s, when there was a surge of interest in Internet companies amid high expectations from startups, most of which turned out to be economically ineffective.

Now analysts also see a tendency to sell previously acquired shares in anticipation of a serious correction in the markets. For example, the Japanese investment fund SoftBank and American entrepreneur Peter Thiel have already gotten rid of the securities of the world's leading manufacturer of graphics processors and Nvidia chips. It is also important to note that according to a Bank of America survey conducted in November, one in two portfolio managers said that the risk of an "AI bubble" could negatively affect the global economy.

Hype on AI

Vladimir Korovkin, Deputy Director of the Center for Communications and Digital Solutions and head of research at the Skolkovo School of Management, noted in an interview with Izvestia that such dynamics of attitudes towards technology were described back in 1995 by Gartner analyst Jackie Fenn in the form of a "hype cycle."

— The essence of it is that at first completely unrealistic expectations are placed on the new technology. Then comes a phase of acute frustration, which is followed by a stage of "enlightenment" (that is, understanding what really works and what doesn't) and reaching a "productivity plateau" when technology turns into just a working tool, he explains.

Now, according to the expert, with regard to generative AI, we have a transition point from the first phase to the second.

— It turned out that he is capable of making mistakes often and very unpleasantly, that the apparent originality of his answers and recommendations is the same for everyone and, therefore, does not give an advantage in competition (and sometimes, on the contrary, worsens the competitive position). This is a normal process of probing the possibilities of technology in different areas of life and the economy," notes Vladimir Korovkin.

It should also be understood, he notes, that the first stage is not final and its onset does not undermine the importance of the major technological breakthroughs made in recent years.

— Therefore, the collapse of the "bubble" is unlikely to have dangerous consequences for anything other than the current capitalization of companies specializing in AI development. Probably, some small players will not survive this stage, but this is a normal market process. Their technologies, which have potential in business, will be picked up by larger companies," explains Vladimir Korovkin.

Georgy Bregman, founder of AI Disraeli, an expert on artificial intelligence and robotics at Skolkovo, adds that concerns about the "AI bubble" have a double foundation: financial and hardware. According to him, on the one hand, we see an overheating of the investment market: in the first quarter, $73.1 billion was poured into AI startups (almost 60% of the total venture capital market).

According to the expert, Sam Altman, CEO of OpenAI, rightly warned that "investors are too excited," and now the market is starting to demand real payback, not just hype.

— On the other hand, and this is the main hidden threat, a tight monopoly on computing power has been formed. The graphics card market, de facto controlled by Nvidia, operates in a bottleneck mode. The lion's share of advanced chips goes to the "American five" (Microsoft, Google, Meta* (the company's activities are recognized as extremist and banned in the Russian Federation), Amazon, Apple), leaving only crumbs to the rest of the world. This creates the risk of a colossal digital stratification, where the development of AI in most countries will be artificially hindered by the lack of hardware," explains Georgy Bregman.

Will AI replace humans

The question of whether AI can replace humans has been actively discussed recently, and the introduction of technology into people's lives is already underway: unmanned trains, disease diagnosis, and writing scientific papers. However, many experts see AI as an assistant that enhances people's capabilities. They note that the functionality will change and transform, but neural networks will not completely replace human labor.

However, working with generative AI and research has also shown that technology can significantly slow down workers by making mistakes. A person has to spend a lot of time searching for and correcting flaws.

Despite what is happening, strategic marketer Alexander Azar believes that people who abandon neural networks risk becoming unclaimed.

— Many people worry that AI will "take away" jobs. But from my point of view, this is not the end, but a new stage in the development of society. Simple tasks will indeed be automated, freeing people up for more meaningful and complex intellectual roles. This is a chance to improve your skills, become more in demand and free in your choice, rather than competing with machines at the level of mindless mechanics," says Alexander Azar.

The expert highlights several advantages of introducing AI into human life. For example, neural networks will allow you to measure and optimize processes that were previously difficult to "digitize." With their help, you can analyze data faster, predict and reduce costs. AI helps to structure thoughts and express ideas, whether it's text, music, or pictures. However, the expert also draws attention to the fact that human critical thinking skills, the ability to verify facts and build arguments can degrade if you rely too much on neural networks.

What is happening with AI in Russia

Georgy Bregman explains that the situation in Russia today is radically different from the Western world.

— There is no speculative bubble in the country, as development follows the path of technological sovereignty. Being under sanctions pressure and without direct access to top Nvidia chips, Russia is giving an asymmetric response. While Western giants solve problems by brute force, "throwing" their capacities, Russian engineers (Yandex, Sberbank, Skolkovo) create optimized neural network architectures that require fewer resources, the expert explains.

There will be no full—scale collapse - there will be a cleansing of the market, the expert believes. Those who switch from text generation to autonomous actions will survive.