Researchers from George Washington University have conducted a groundbreaking experiment by replacing policymakers at a Federal Reserve meeting with AI agents. The simulation, designed to mimic the July 2025 Federal Open Market Committee (FOMC) session, demonstrated how political influence can deeply affect monetary policy debates, even within institutions built on formal rules and independence.
To create a realistic model, scientists trained AI assistants on the historical stances, biographies, and public speeches of actual policymakers. These agents were then tasked with analyzing real-time economic data and financial news to decide on interest rate adjustments. The results were striking: under simulated political pressure, the AI agents became increasingly polarized, producing a wider spread of dissenting opinions. This highlighted a key insight — that even a body as structured and rule-based as the Fed is not entirely immune to external political forces.
The study suggests that while AI can offer analytical precision and data-driven insights, it also reflects the biases and conflicts inherent in the information it processes. In other words, AI systems may replicate not only human decision-making strengths but also human vulnerabilities when it comes to partisan divides.
The experiment ties into a broader global trend where central banks are cautiously adopting AI technologies. While few institutions are ready to entrust monetary policy decisions entirely to machines, many are already deploying AI to support economic research and forecasting. For example, the European Central Bank applies machine learning to forecast inflation across the eurozone. The Bank of Japan leverages AI for information gathering and deep economic analysis. Meanwhile, the Reserve Bank of Australia is experimenting with tools that generate concise policy summaries, enhancing staff efficiency without replacing human judgment.
Importantly, central banks emphasize that AI is not a decision-maker but a tool for improving efficiency. Michelle Bullock, Governor of the Reserve Bank of Australia, recently explained that AI is used to enhance research, streamline analysis, and support staff contributions, not to dictate monetary or financial policy.
According to an April report from the Bank for International Settlements, most central banks remain in the early stages of AI adoption but recognize its strategic potential. From analyzing vast amounts of data to generating actionable insights, AI could play a pivotal role in shaping how institutions monitor global markets and assess risks.
Conclusion: The simulation of the Fed’s deliberations with AI agents underscores both the promise and limits of artificial intelligence in monetary policy. While it can enrich analysis and highlight hidden dynamics, it cannot fully escape the political realities that shape economic governance. As central banks continue experimenting, the key challenge will be balancing AI-driven efficiency with the human oversight needed to preserve institutional trust and independence.
