What happened
This IMF Working Paper by Patrick A. Imam and Jonathan R.W. Temple, drawing on 1,097 comparable country-period transitions, finds that 'the average time required for economies to leave the lowest [productivity] state was 16.1 years for capital intensity, 31.5 years for measured human capital and 44 years for total factor productivity (TFP)' — showing capital and schooling diffuse far more readily across countries than productive capacity itself. Using illustrative counterfactual scenarios, the paper estimates that broad, widely-shared AI-enabled technology diffusion could cut the average escape time from the lowest-productivity category to 28.6 years, while concentrated/unequal AI diffusion could push it to 58.8 years, with over half of economies still trapped after a century. The authors argue AI narrows global development gaps only via a third channel — lowering the cost of adapting and applying technical knowledge locally — not simply by raising frontier productivity or disproportionately rewarding already-skilled workers.
Why it matters
Provides government and multinational strategy teams a quantitative framework for judging whether AI adoption programs will narrow or widen international productivity gaps, clarifying that AI access alone will not close development gaps without complementary investment in skills, institutions, and data.
Action needed
Apply the paper's three-pathway framework (frontier-push, skill-complementary, knowledge-diffusion) when evaluating whether AI deployment strategy in target/emerging markets favors convergence or divergence.