A world-renowned economics scholar warned that an era is coming when artificial intelligence (AI) reads the direction of interest rate decisions before the Central Bank does and makes money ahead of people.
Reuters said on the 31st (local time) that Markus Brunnermeier, a professor of economics at Princeton University, presented a paper on how AI will shake up finance at the Jackson Hole Economic Policy Symposium (Jackson Hole meeting) in Jackson Hole, Wyo., on the 29th. Professor Brunnermeier is a world-renowned scholar who has led the Bendheim Center for Finance at Princeton and studied asset bubbles, liquidity, and financial stability. He has been involved in financial policy around the world to the extent that he advised the International Monetary Fund (IMF), the Federal Reserve Bank of New York, and the European Systemic Risk Board. His new paper contained almost none of the equations that typically fill economics papers. Instead, sentences closer to philosophy took their place.
The Jackson Hole symposium is a global economic event hosted annually since 1978 by the Federal Reserve Bank of Kansas City, now in its 49th year. This year's theme was "Implications of financial innovation for payments and policy." In line with the theme, papers mainly examining how stablecoins and distributed ledger technology will change the Central Bank and the real economy shared the stage. Distributed ledger technology refers to digital ledger-related technology in which network participants, without a central server or administrator, jointly verify transaction records and store them in a distributed manner.
Brunnermeier's presentation was the final session of the symposium. He argued that because AI far outstrips people in processing information, it will be able to know with near certainty whether interest rates will be raised or lowered before the Central Bank even reaches its own conclusion. He called this information gap between AI and the Central Bank "asymmetric understanding."
Brunnermeier predicted that AI will reach conclusions on the direction of financial policy ahead of people and devise strategies to evade regulation or front-run to profit using this information. He said such AI does not distinguish whether a money-making method is legal or not. In the paper, he wrote, "In a world where humans and AI agents are mixed in transactions, extreme situations can arise in which markets become more capricious than they are now," and "From the Central Bank's standpoint, it will have to play a game against AI that understands the policy direction better than the Central Bank itself."
Brunnermeier said that for the Central Bank to avoid being used by AI or falling behind, it should use language with ambiguous intent so that AI will find it hard to understand or predict. The warning implies that transparency could instead become a pretext to shake markets. This prescription runs counter to the last 30 years of monetary policy trends. Over the past 30-plus years, Central Banks have incrementally increased the information they disclose to markets.
The Fed began in February 1994 to disclose why it made policy decisions. Six years later, starting in 2000, it issued statements even after meetings where it did not change interest rates. In 2011, then-Fed Chair Ben Bernanke held the first news conference immediately after a Federal Open Market Committee (FOMC) meeting. The following year, Bernanke put the Fed's inflation target into writing. The Bank of Korea also immediately issues a decision statement when the the Bank of Korea's monetary policy committee sets the base rate, and the governor explains the background at a press briefing before releasing the minutes. This trend of disclosure has been credited with increasing investor understanding of previously closed Central Bank policy directions, reducing market volatility, and making transactions more efficient.
Brunnermeier warned, "Unbridled AI undermines trust in institutions," and "Those without sophisticated policy tools will be pushed into a disadvantageous position." To overcome this, he proposed holding two news conferences—one to convey concrete criteria to people and another to provide analysis to be used as machine-learning data—and separately managing AI agents like financial influencers. He said, "If policy transparency turns into predictability, public authorities can become trapped in financial markets."
However, major outlets reported that Central Bank officials gathered at Jackson Hole did not pay much attention to this pessimistic scenario. In a keynote speech on the 28th, Chair Waller offered optimism on AI that differed from Brunnermeier's. Waller said, "We are embracing AI as a new variable, perhaps a new factor of production," and "We think this variable will leave some kind of imprint on both the economy and the operation of monetary policy." However, Waller did not specify what form that impact would take.