Central Banks & AI Watch

Two consequential developments emerged as the IMF–World Bank Annual Meetings opened in Bangkok. First, the Financial Stability Board brought its finalised AI supervisory practices to the G20, moving international coordination beyond risk identification toward a common framework for responsible adoption.

Second, the IMF quantified an increasingly important divergence: AI infrastructure investment is supporting global growth near 3%, but its benefits are concentrated geographically while its effects on employment grow more disruptive.

On October 12, FSB Chair and Bank of England Governor Andrew Bailey presented the organisation's latest financial-stability agenda ahead of the October 15 G20 meeting. The FSB confirmed it has finalised sound practices for responsible AI adoption, incorporating public consultation feedback. The framework addresses organisation-wide governance, AI lifecycle management, human oversight, cybersecurity and third-party dependencies.

These are nonbinding supervisory practices, not a new international regulation — but their finalisation moves AI supervision toward a common international reference framework. Bailey's letter also connects AI vulnerabilities with volatility in sovereign bond markets and fragilities in private credit.

The second-order effect is a supervisory paradox. Convergence could reduce fragmentation and strengthen resilience, but compliance costs could favour large institutions with sufficient technology, capital and specialist personnel. Stronger institutional AI controls could coexist with greater concentration among the technology providers on which those institutions depend. For central banks, the question becomes whether better individual-bank governance is sufficient when many institutions share the same models, infrastructure and service providers.

Speaking in Bangkok, IMF Managing Director Kristalina Georgieva said AI infrastructure investment is helping sustain global growth at approximately 3%, despite the negative effects of the energy supply shock. AI-related investment could contribute approximately 0.5–1 percentage point to growth in advanced Asian economies and 0.2–0.8 points in emerging Asian economies — estimated contributions, not confirmed realised gains.

The monetary implication is that central banks must distinguish three effects: capital formation (higher investment demand, potential inflationary pressure); productivity diffusion (higher potential output, possible disinflationary consequences); and geographical divergence (different outcomes depending on each economy's position in the AI supply chain). The Great Split therefore creates different monetary-policy conditions across countries exposed to the same transformation. Today's remarks establish no new rate decision — they strengthen the case for separating the investment phase from the productivity phase.

President Tharman Shanmugaratnam identified another emerging risk: AI is already reducing demand for some entry-level graduate positions in highly exposed sectors. When graduates lose access to first professional positions, they may also lose opportunities to acquire experience and advance through conventional career structures. Tharman warned that insufficient policy responses could produce a new divergence between advanced and developing economies.

This complicates the labour-market problem for central banks. AI can support investment and output while weakening employment in specific occupations — and the resulting unemployment may not respond to conventional monetary stimulus if it reflects structural displacement rather than insufficient demand. This is an interpretation of the emerging evidence, not a formal change in the MAS reaction function.

The United States and major Asian economies are capturing substantial portions of AI infrastructure investment. European companies remain more dependent on internally generated funds and less able to mobilise risk capital at comparable scale. Meanwhile, global AI financing demand can lift long-term interest rates, affecting European investment even where domestic AI capital formation remains limited.

Europe faces a threefold challenge: global capital-cost pressure, uneven AI productivity diffusion, and emerging employment displacement. The response extends beyond monetary policy. EU policymakers need to accelerate industrial AI adoption, improve access to growth capital, expand reliable energy capacity and redesign workforce-transition mechanisms — and ensure AI supervision encourages competition and resilience rather than reinforcing dependence on a small number of foreign technology providers.

On October 13, ECB Chief Economist Philip Lane is scheduled to speak ahead of the International AI Summit in Dublin; his remarks may show how the ECB is integrating AI investment and productivity expectations into monetary analysis. On October 15, G20 finance ministers and central-bank governors will discuss the FSB's supervisory agenda. The IMF's updated global economic outlook will provide an opportunity to evaluate whether AI-related investment is changing growth, inflation and financial-stability assessments.

Five Key Takeaways

For investors and policymakers

• Financial supervision: the FSB's finalised AI practices mark a meaningful advance toward international supervisory coordination.
• Growth divergence: IMF estimates suggest AI capital formation is creating materially different growth opportunities across Asian economies.
• Employment: AI-related displacement may emerge before productivity benefits are broadly distributed, complicating assessments of labour-market slack.
• Investment strategy: distinguish companies capturing measurable productivity returns from those merely increasing AI expenditure.
• European competitiveness: accelerate the conversion of savings, energy and adoption into productive investment while preserving financial resilience.

Economic growth generated by AI investment is not equivalent to productivity growth generated by AI adoption. The next monetary regime will not be determined by the scale of AI investment alone, but by how quickly that investment translates into productive capacity, financial returns and broadly distributed economic gains.