AI could help developing countries reduce inequality, but only if basics are in place, World Bank says

TEXT START: Generative AI could enhance more than 16% of jobs in developing economies, while 14.2% of jobs in high-income countries face automation risk, compared with 4.5% in low- and middle-income nations

The Dissection

The article converts an obsolescence problem into a development-readiness problem. It argues that poorer countries can use AI to amplify workers, improve agriculture and public services, and reduce inequality—provided they acquire electricity, connectivity, data, skills and institutions first.

Its real function is to preserve the developmentalist storyline: build the basics, customize the tools, then advance toward frontier AI. The article acknowledges concentration and inequality, but treats them as policy defects around the system rather than consequences of AI capital ownership.

The Core Fallacy

It treats “augmentation” as a durable economic destination rather than a transitional phase.

The comparison between 16% of jobs potentially enhanced and 4.5% exposed to automation is also structurally misleading. Low exposure in developing countries largely reflects low digitization and weak infrastructure—not durable protection from substitution. Once AI becomes cheaper, more capable and embedded in local systems, the same lag that limits augmentation can accelerate displacement.

Productivity gains do not automatically become wages, bargaining power or mass employment. Under the Discontinuity Thesis, the decisive question is who owns and controls the AI systems, distribution channels, data and energy. If those remain concentrated, “amplification” means higher output with fewer economically necessary workers.

Hidden Assumptions

• Existing jobs will remain necessary after AI raises productivity.
• Productivity gains will flow to workers rather than AI capital owners.
• Infrastructure can be built quickly enough to outrun competitive automation.
• Governments can coordinate stable human-only economic domains at scale.
• Local customization will create local bargaining power instead of dependence on foreign platforms.
• More growth will translate into broad-based income and reduced inequality.
• The gap in AI capability is primarily a lack of basics, rather than a structural ownership gap.
• “Frontier AI” will remain a tool that complements labor instead of becoming the superior substitute.

These assumptions are not demonstrated. They are the load-bearing fiction of the article.

Social Function

Primarily transition management, with a partial truth wrapped in ideological anesthetic.

The infrastructure warnings are real: countries without electricity, connectivity, data and institutional capacity will capture less value. But the framing reassures policymakers that catching up technologically can preserve participation. It shifts attention from the death of the wage-consumption circuit to an administrative checklist and allows owners of AI capital to present concentration as an implementation problem.

The Verdict

This is not a rebuttal to obsolescence. It is a deployment memo for countries arriving late to the automation race.

Developing countries may receive a temporary augmentation window because their economies are less digitized. That is lag, not immunity. If they build the basics without securing ownership and control of AI capital, they may produce more while employing fewer people and exporting the surplus. The basics can determine who manages the carcass; they cannot, by themselves, resurrect the mass-labor system.