A United Nations report warns that the spread of artificial intelligence may deepen global inequality as adoption and investment accelerate unevenly across the world. The preliminary analysis from the independent international scientific panel on AI - the first global scientific body of its kind - offers both a diagnosis of the risks and a shared framework for responsible development.
The panel, comprising 40 independent experts established by the UN General Assembly last year, structured the report as a toolkit for member states. "Access to AI tools alone does not produce equal benefit," the report says. Countries that depend on foreign models, cloud infrastructure, and data pipelines may gain access while losing practical control over standards, safeguards, and local fit.
A toolkit for responsible AI adoption
The document provides initial guidance on harnessing AI's potential across industries while addressing threats. Its recommendations include:
- Developing local AI infrastructure, such as data centers
- Improving AI literacy in schools and the workforce
- Investing in local developer talent
- Building AI safety institutes
- Creating strategies to combat disinformation
- Continuously measuring how AI systems perform after release, with real users and tasks
The panel also acknowledges the environmental costs of data centers - high energy and water consumption and potential greenhouse gas emissions - and notes that securing a reliable energy supply is a prerequisite for attracting investment in computing infrastructure.
Concentration and the digital divide
While more than a billion people now use AI weekly, adoption in the global south lags far behind the global north. The US and China dominate the development of leading models and investment in compute hardware. "The concentration of AI capabilities in a small number of firms and countries could enable authoritarian capture and undermine democratic accountability," the report states. Most nations, including advanced economies, lack the technical expertise to assess frontier models or participate meaningfully in their governance.
Language barriers compound the gap. Generative AI tools perform well in English and other widely used languages, but many languages are either excluded or suffer much lower performance. The report cites a medical translation error where a Tigrinya output confused smallpox with syphilis, gonorrhoea as diabetes, and "intravenous antibiotics" as "intravenous insecticides." "These mistranslations can be life-threatening," the report notes. Beyond language, 2 billion people - almost a third of the global population - remain completely offline, according to the International Telecommunication Union.
Why this matters for government and IT professionals
For government policymakers, the report's suggestions offer an early template for national AI policy and safety institutes - concrete steps to strengthen democratic accountability. The emphasis on building local computing infrastructure and developer training aligns with the priorities of AI for IT & Development professionals, who will be central to these efforts. The call to continuously measure AI systems after release also points to a growing need for monitoring and evaluation skills in both the public and private sectors.
Your membership also unlocks:
