Artificial intelligence (AI) is rapidly transforming financial markets, but experts warn of potential hidden systemic risks associated with this technological revolution. The current AI euphoria in the stock market, fueled by companies like NVIDIA developing powerful machine learning processors, may obscure a more troubling reality. While AI holds the promise of revolutionizing trading and risk management, it could paradoxically render our financial systems more fragile and susceptible to catastrophic failures.
Renowned author Jim Rickards, in his new book Money GPT, highlights the widespread adoption of AI in financial markets, with major investment banks on Wall Street implementing it. However, Rickards controversially asserts that this widespread adoption could amplify market crashes beyond anything witnessed before. He introduces the concept of the “fallacy of composition,” where actions that make sense for individual market participants could lead to disaster when adopted by everyone. This analogy is illustrated by the scenario of one fan standing up at a football game, which works to provide a better view for that individual but becomes problematic when everyone follows suit, resulting in an obstructed view for all.
The phenomenon of the fallacy of composition could manifest during market downturns. While it may be prudent for individual investors to sell during a crash, if AI systems controlling significant capital all execute similar strategies simultaneously, the outcome could be catastrophic. Rickards also points out that removing human judgment from the equation poses a significant risk. The historic role of specialists on the New York Stock Exchange, responsible for maintaining orderly markets, is highlighted. These specialists were tasked with equilibrating the market during waves of sellers, a nuanced judgment that Rickards argues is lacking in today’s AI systems.
AI introduces unprecedented risks through its speed and synchronicity. Market panics are not new, but the automated nature of AI-driven trading can accelerate market movements and create feedback loops that human traders might otherwise interrupt. Rickards warns of the speed, amplifying effect, and recursive function of these events. The concerns extend beyond stock markets to the banking system itself, as evidenced by the recent collapse of Silicon Valley Bank, which occurred rapidly due to digital technology.
Rickards suggests that abandoning AI entirely is not the solution. Instead, he advocates for the implementation of more sophisticated circuit breakers and regulatory frameworks. He proposes “cybernetic” approaches that gradually slow market activity during periods of stress, rather than implementing sudden stops.