Panel discussion on AI and the resilience gap, focused on how AI diffusion may widen or narrow income gaps across countries, and what policy makers should do to prepare.
Watch on YouTube ›Get the market thesis, key claims, assets, contradictions, and follow-up questions from any financial video — then unlock a version personalized to your portfolio, watchlist, and favorite speakers.
This IMF panel examined AI as a macroeconomic and policy issue rather than a stock or market-trading setup. The conversation centered on uneven AI diffusion, the gap between advanced and low-income economies in preparedness and access, and the risk that AI could increase divergence instead of convergence. Speakers repeatedly emphasized that the gains from AI will depend less on raw model capability and more on complementary investments: digital infrastructure, data access, governance, workflow redesign, workforce skills, and social protection. Peter McCroy of Anthropic described the Anthropic Economic Index, saying usage of Claude has been tracked across more than 150 countries and that early adoption is geographically concentrated, with the US and Singapore showing far higher per-capita usage than expected. …
Near term, the actionable setup is policy uncertainty: governments and firms should treat AI as a rapidly moving but uneven diffusion story, with the biggest tactical risk being under-preparedness on data, workflows, and workforce transition rather than a clean economy-wide boom.
Over the next few quarters, expect a gradual-but-uneven adoption path: pilots will expand, a few structured functions may scale quickly, and tradable services plus routine knowledge work may show the first strain. The base case improves if institutions add wage insurance, retraining, and stronger digital infrastructure; it weakens if diffusion outpaces policy.
Structurally, AI looks like a general-purpose technology that can raise trend productivity while also widening international and within-country gaps if complementary investments are missing. The durable question is which economies build the data, governance, and learning systems that let them capture the gains without locking in dependence or divergence.
Claude usage is geographically concentrated, with the US and Singapore showing much higher per-capita usage than expected from population share.
Peter says the index tracks usage across 150+ countries and notes US and Singapore usage per capita are several times higher than expected.
AI could widen divergence between advanced and low-income countries unless policy intervenes.
Bo explicitly says there is real risk of divergence rather than convergence because of AI and that policy intervention is needed.
The productivity gain from AI may range from 0.1% to 0.8% annually, with advanced economies benefiting roughly twice as much as low-income countries.
Bo gives the range and says gains can be twice as high in advanced economies.
Anthropic has taken the world by storm. You're tracking usage of the technology across economies through the Anthropic Economic Index. What patterns have you identified?
Peter McCroy explains that the Anthropic Economic Index tracks how people and businesses use Claude across 150+ countries. The most striking finding is that early adoption is geographically concentrated — the US has over 4x expected usage per capita based on working-age population share, and Singapore has about 5x. This pattern correlates strongly with GDP differences between countries, raising questions about whether benefits accrue to already-rich countries.
One of the biggest risks of AI is that it could drive a deeper wedge between economies, fueling income divergence. How concerned are you?
Bolie sees a real danger of divergence but also reasons for hope. On the risk side, AI could produce annual productivity gains of 0.1-0.8%, but advanced economies' potential growth benefit can be twice as high as low-income countries. Without proactive policy intervention, divergence could widen across three dimensions: AI exposure, preparedness, and access. On the hopeful side, competition between AI models (open vs closed, US vs China vs European tech) can drive down prices, and AI could level differences — e.g., deploying AI at scale in developing countries to address shortages of doctors and teachers could make it a leveler instead of a divider.
What should policymakers be focused on at this point regarding AI?
Bolie recommends two areas: First, AI preparedness — countries should strengthen digital infrastructure, upskill labor, establish robust governance with guardrails. The IMF's AI Preparedness Index shows advanced economies have advantages in infrastructure, labor skills, and governance that developing countries need to catch up on. Second, countries should consider the macro implications of AI — scenario planning, risk management, and stress-testing their macro frameworks to manage potential disruption.
Unlock the full claims, asset map, scores, related transcripts, follow-up questions, and AI chat — shaped around your portfolio, watchlist, favorite speakers, and risks.
