The uneven transition
Bars, colour and label follow the selected axis. Green means lower pressure or stronger capacity; red means higher pressure or weaker capacity.
The inverted map
Advanced economies are most exposed to AI, but also best equipped to use the technology and absorb the shock. Low-income countries are less directly exposed, but have weaker protection and less ability to capture the gains.
Demography reverses the risk
Where labour is shrinking, AI can solve a shortage. Where labour is growing, the same technology can make entry-level jobs scarcer. The highest social risk therefore does not necessarily sit where exposure is highest.
Two kinds of preparedness
AIPI is useful as a measure of AI adoption and upside capacity. It should not be read as a measure of social resilience. Safety nets, work culture, local labour mobility and institutional trust must be judged separately.
Three scenarios
Acceleration brings rapid capability growth and high divergence risk. Friction brings blocs, export rules and uneven access. Sharing requires access, investment and gains to spread widely enough that the development ladder is not pulled up.
Sources: IMF work on AI exposure and AIPI, NBER on premature deindustrialisation and the China shock, IEA on energy and AI, and Federal Register/AP on the AI Diffusion Rule.