World Bank Sees AI Lifting 16.2 Percent of Jobs but Warns of Wider Gaps
Artificial intelligence could meaningfully raise productivity in sixteen point two percent of jobs in developing economies, close to the eighteen point seven percent expected in high income countries, while putting a far smaller share of jobs at risk of automation, according to the World Bank Group’s World Development Report 2026, released on 4 August.
The asymmetry is the report’s central finding. Four point five percent of existing jobs in low and middle income economies are assessed as at risk of automation by generative artificial intelligence, against fourteen point two percent in high income countries. Jobs in richer economies are therefore more than three times as likely to be exposed. The Bank attributes the gap to employment structure: developing economies hold more workers in manual and less cognitively intensive work, while advanced economies carry larger shares in the administrative, analytical and knowledge based tasks that generative systems can increasingly perform.
| Share of jobs | Developing economies | High income economies |
| Productivity meaningfully boosted | 16.2 percent | 18.7 percent |
| At risk of generative AI automation | 4.5 percent | 14.2 percent |
| Ratio of upside to exposure | About 3.6 times | About 1.3 times |
The ratio in the final row is arithmetic on the Bank’s two figures rather than a figure the Bank itself publishes. It shows that in developing economies the share of jobs that could gain is roughly three and a half times the share exposed, while in high income economies the two are close to balanced.
Indermit Gill, Senior Vice President and Chief Economist of the World Bank Group, said artificial intelligence had thrown developing economies a lifeline and they should seize it. He added that they do not need large models or big data centres to benefit, and that by adapting small, low cost tools to local conditions they can bring better medical care, education, judicial services and agricultural extension within reach of millions. He also warned that they must hurry, because the technology is spreading faster and is more context specific than earlier general purpose technologies such as electricity and the internet.
The speed comparison is stark. The Bank calculates that it took about eighty years for the steam engine to reach lower income countries and forty years for electricity, and twenty years for the internet. Middle income countries accounted for half of ChatGPT’s global traffic within six months of its launch.
The report lands at a difficult moment. Developing economies are in their weakest average growth performance in three decades, and the Bank argues artificial intelligence could lift that performance before the end of the decade. It sets out a three step path, adopt available tools, adapt them to local conditions, and over time advance toward frontier development, with each step requiring more investment, skills and infrastructure than the last. Adapting, rather than merely adopting, is where the Bank expects the largest benefits, because tools trained in high income countries may produce recommendations misaligned with local norms.
Against that, the Bank is explicit about what could go wrong. It warns that without deliberate action artificial intelligence could widen gaps between countries, increase inequality within them, concentrate market power, weaken trust in public institutions and create new risks for safety, rights and social cohesion. It notes that a small number of companies in a few economies control the most advanced models, the chips they rely on and the data centres that run them, which creates dependency risk, though it also allows countries to customise existing models without spending billions building their own.
| The case for | The case against |
| 16.2 percent of jobs could gain in productivity, near the high income share | Complements are missing: electricity, connectivity, computing, data, skills, institutions |
| Automation exposure is a third of the high income level, at 4.5 percent | Nearly a third of rural schools in Sub-Saharan Africa lack reliable electricity and more than two thirds lack dependable internet |
| Diffusion is far faster than steam, electricity or the internet | Model, chip and data centre production is concentrated in a few firms and economies |
| Adoption needs no large models or data centres | Tools trained elsewhere may not fit local languages, norms or institutions |
| Concentration lets countries customise rather than build from scratch | Frontier development is unrealistic for most countries in the near term |
| Public services stand to gain: tax collection, social programmes, disaster response, health, education | Risks to safety, rights, social cohesion, market competition and public trust |
Infrastructure is where the Bank places the binding constraint. In Sub-Saharan Africa nearly a third of rural schools still lack reliable electricity and more than two thirds lack dependable internet access. The Bank is working with partners through Mission 300 to bring energy access to three hundred million people across the region by 2030. It also calls for wider access to computing power and for better local data, including in local languages, so tools can be tailored rather than imported wholesale.
On governance, the report suggests governments start with voluntary industry standards, anchored in international cooperation to avoid regulatory fragmentation, and apply existing law where voluntary measures fall short. Gaurav Nayyar, the report’s Director, said the window to get this right is narrow, and that countries which build the foundations now, power, connectivity, skills and institutions, will be positioned to adopt and adapt the technology for their people.
Why it matters: The headline finding is easy to misread as reassurance. A four point five percent automation exposure against sixteen point two percent productivity upside sounds like developing economies have drawn the better hand, and on employment structure alone they have. But the same structural features that limit their exposure, fewer knowledge workers and less administrative employment, are what make the productivity gain hard to capture, because capturing it requires exactly the electricity, connectivity, data and institutional capacity the Bank identifies as missing. The two figures are not independent. The low automation risk and the difficulty of realising the upside come from the same source, which is why the report reads less as a forecast than as a conditional.
Looking ahead: The Bank’s three step sequence puts the near term test on adoption and adaptation rather than on frontier development, so the measurable signals will be procurement and evaluation frameworks, local language data, and whether pilot projects are assessed rigorously enough to show which ones work. Mission 300’s progress toward three hundred million people by 2030 is the clearest quantitative marker of whether the analog foundations are being laid.
Sources: World Bank Group, World Development Report 2026: The Promise of Artificial Intelligence; World Bank Group press release, 4 August 2026.
