The Bank for International Settlements (BIS) identified a collapse of the artificial intelligence (AI) bubble, alongside inflation and fiscal stress, as one of the three major threats to the global economy in its annual report released Sunday. The BIS warned that "disappointing returns (on AI) could trigger a sudden withdrawal of funds, and the current capital expenditure boom could turn into a prolonged investment slump, with knock-on effects on financial conditions." In other words, if the "circular financing" loop—seen as a vulnerability in the AI industry—were to break, it could trigger a recession similar to the 2008 financial crisis.
The term "AI Ouroboros" is often used in the AI industry. It derives from the Ouroboros, a serpent from ancient Greek mythology that endlessly regenerates itself while biting and swallowing its own tail. Initially, the term was used to warn of the dangers when AI relearns from content generated by AI. The logic is that if such recursive learning repeats, distortions and false information accumulate, accuracy and reliability decline, and this could lead to the collapse of AI learning models.
These days, the term is also used to point out the risks of an AI ecosystem in which investment and revenue circulate, intertwined and interdependent. For example, when Big Tech companies acquire stakes in AI ventures and startups, those companies in turn use the Big Tech firms' cloud services or purchase high-performance AI chips from the likes of Nvidia. In this process, both sides see their revenue and corporate value grow, and venture investors reap enormous profits.
The problem, as the BIS points out, is that if the performance of mega-scale cloud providers such as Amazon and Google falls short of expectations, investment funds could plummet, putting the entire AI industry at risk. It is similar to how, in the Battle of Red Cliffs during China's Three Kingdoms period, Cao Cao chained his ships together in a connected formation, only to have the entire Wei army sink under the Wu kingdom's fire attack. To be sure, the prevailing forecast is that AI-related investment will continue for the time being, exerting a positive influence on innovation and growth. Yet there are also concerns that if the AI bubble bursts, the shock could be greater than that of the dot-com bubble in the 1990s. Optimism should be maintained, but full preparations are called for.
