AI impacts on developing nations gauged - World - Chinadaily.com.cn

TEXT START: While developing economies face lower job displacement risks from artificial intelligence compared with developed nations, gaps in infrastructure development, institutional capacity and relevant skill sets largely restrict their access to the benefits that AI has to offer, said a recent World Bank report.

The Dissection

The article performs a controlled reframing. It converts lower immediate AI exposure into apparent strategic advantage, then presents infrastructure, skills and institutional weakness as solvable prerequisites for entering the AI economy. Its three-step strategy—adopt, adapt, advance—is an adoption roadmap, not a theory of who will own the resulting productivity.

The article contains a real observation: developing economies have fewer jobs currently exposed to generative AI. But that is partly because they have fewer concentrated pools of formal, cognitive and digitally mediated work to automate. Low exposure is not protection. It can signify low productivity, weak bargaining power and exclusion from the sectors where AI-generated value accumulates.

The Core Fallacy

The central error is treating delayed displacement as durable viability.

The article assumes that if AI raises productivity in developing economies, the gains will diffuse into employment, wages, public capacity and broad prosperity. Under the Discontinuity Thesis, that transmission mechanism is precisely what AI destroys. Once cognitive automation becomes cheaper and better, productivity gains accrue primarily to the owners and controllers of AI systems, infrastructure, data, energy and logistics—not automatically to the workers who use the tools.

The report describes AI as a lifeline because it can compensate for shortages of doctors, teachers, administrators and skilled workers. The harsher interpretation is that AI becomes a substitute for building a large domestic professional class. Developing economies may receive useful services while losing the labor-based route to mass productive participation. They become consumers and operating environments for imported intelligence, not owners of the capital producing it.

P1 makes the labor advantage temporary. P2 makes stable human-only economic niches difficult to preserve. P3 then breaks the employment-to-wage-to-consumption circuit. The article measures exposure to the first wave while largely avoiding the terminal mechanism.

Hidden Assumptions

• Productivity improvements will become jobs or higher wages rather than profits, rents and vendor fees.
• Adoption by governments and businesses means meaningful local control.
• Imported systems will transfer capability instead of creating permanent technological dependence.
• Small, low-cost tools will remain affordable and strategically neutral as vendors consolidate power.
• Weak states can regulate dominant AI suppliers without becoming captive to them.
• Infrastructure investment can outrun the speed of AI capability improvement and capital concentration.
• Local adaptation will preserve sovereignty even when models, compute, data pipelines and standards are externally owned.
• A population that is not yet exposed to automation remains economically secure.
• Expanded access to AI-mediated healthcare, education or justice is equivalent to productive participation.
• The frontier will remain open long enough for developing economies to climb toward ownership rather than remain downstream users.

These assumptions conceal the ownership question. The article asks whether developing countries can use AI. The decisive question is whether they can own and control the systems that make human labor economically nonessential.

Social Function

Primary classification: transition management wrapped in ideological anesthetic, with a substantial partial truth.

The partial truth is that infrastructure, electricity, connectivity, skills and institutional competence materially determine who can deploy AI. The anesthetic is the language of a “lifeline” and a narrow policy window. It encourages governments to prepare for adoption while leaving the distribution of ownership, rents and bargaining power mostly untouched.

Its practical function is to manage the transition politically: build foundations, import tools, localize applications and hope productivity arrives before dependency hardens. The vendor-dependence warning is important, but it is treated as a governance risk rather than evidence that the productive core may sit permanently outside the country.

The result is a respectable development narrative applied to a discontinuity problem. It tells states how to become efficient users of an economic system whose commanding assets may belong to foreign firms and foreign sovereigns.

The Verdict

The article correctly identifies a lag, then mislabels it as a lifeline. Developing economies face less immediate job displacement because less of their existing work is exposed—not because they are structurally safer. Their weak infrastructure may delay automation, but it also prevents them from capturing AI rents when automation arrives.

The likely outcome is a double trap: imported AI raises service capacity while imported ownership captures the surplus. Some states and firms can become Sovereigns; a smaller number of workers can survive as indispensable Servitors. The majority receive access to AI-mediated services without gaining control over the productive system. This is not the rescue of mass-employment capitalism. It is preparation for its uneven, externally owned replacement.