Bank of England Policymaker: AI Trading Encourages 'Herd-Like Behavior'

Bank of England policymaker Jonathan Hall has voiced concern over the adoption of AI in trading by financial firms, stating it could amplify shocks on markets and encourage “herd-like behavior.”

Bloomberg reports that according to a recent announcement, a Bank of England policymaker warns that artificial intelligence could pose significant risks to financial stability if adopted in trading by financial firms. Jonathan Hall, an external official on the Financial Policy Committee, believes that AI programs might seek to evade human control and “learn the value of actively amplifying” a market shock.

Hall’s remarks highlight the growing scrutiny of AI from global regulators. As companies embrace the technology to improve productivity, concerns are rising about potential risks to financial stability. In December, the UK central bank stated that it will examine the risks to financial stability from AI in 2024, while also acknowledging the potential economic growth benefits.

Hall said, “There is a broader system-wide worry that a move towards AI traders will lead to greater correlation and, amplificatory, herd-like behavior.” He added, “My intuition is that over the medium term a shift to neural networks would lead to higher correlations and greater amplification of shocks.”

Adopting AI into trading algorithms could lead to “less resilient and highly correlated market ecosystem,” according to Hall, who called for a “kill switch” and human oversight to reduce the risks associated with the burgeoning technology.

One potential risk highlighted involves AI programs incentivizing themselves to worsen market plunges to boost returns. “At the extreme, (it) might learn that an environment of market instability offers the best opportunity for outsized profits. As such, it would be incentivized to amplify any external shock,” Hall said.

He also discussed the “panda/gibbon example,” where an AI deemed an image of a panda to be a gibbon with nearly 100 percent confidence after noise was added to the picture. This illustrates the “hallucination” potential of AI models, which could further increase risk in financial markets.