The UN warns that AI advancements may increase global inequality, highlighting risks of concentration in technology and access disparities.
The United Nations has issued a stark warning regarding the rapid advancement of artificial intelligence, suggesting it may exacerbate global inequality. This caution comes from a preliminary report compiled by the UN's Independent International Scientific Panel on AI, which includes insights from 40 scientists and experts.
The report emphasizes that mere access to AI technologies does not equate to equitable benefits across nations. Countries that depend on foreign models and cloud services may find themselves with access to advanced technologies but lacking control over their standards and local adaptations.
One of the critical issues highlighted is the concentration of computational resources and the development of sophisticated AI models. The findings reveal that the United States possesses approximately 75% of the computational power of the world’s leading supercomputers, while China holds around 15%. Together, these nations account for nearly 90% of the computations necessary for training the most advanced AI systems.
The UN warns that this concentration of AI capabilities in the hands of a few companies and nations could bolster authoritarian regimes and diminish democratic accountability. To address this, the report suggests that nations lagging in AI development must invest in their own computational infrastructure and nurture local talent.
“In the long run, the gap between rapidly growing functional capabilities and the availability of effective risk management methods could lead to catastrophic outcomes,” the report cautions.
Additionally, the report addresses linguistic disparities, noting that generative models tend to perform better in English and other major languages, often yielding poor results in less commonly spoken languages. For instance, it cites a translation error in Tigrinya that could have significant implications for health outcomes.
“Such translation errors can pose life-threatening risks,” the authors stated.
Access to the internet is another contributing factor to inequality, with the International Telecommunication Union estimating that 2.2 billion people will remain offline by 2025. For these communities, the deployment of AI is contingent upon access to the internet, energy, and digital literacy.
The preliminary report is not a regulatory framework but offers a scientific evaluation of the risks and opportunities presented by AI for governments to consider in policy-making. Key recommendations include:
- Establishing local AI infrastructure and data centers;
- Enhancing AI education in schools and workplaces;
- Investing in developers and technical skills;
- Creating institutions focused on AI safety;
- Formulating strategies to counter disinformation;
- Evaluating AI systems post-deployment in real-world scenarios.
Moreover, the report flags potential risks associated with disinformation, fraud, election interference, human rights abuses, and impacts on labor markets. The Bank for International Settlements recently noted that the AI investment surge, while bolstering the global economy, is also generating macro-financial risks.
