The rapid development of artificial intelligence (AI) is ahead of governments' ability to regulate it, the United Nations has warned, stressing that the opportunity to develop an effective global framework for the governance of this technology will not remain open forever.
اضافة اعلان
This came in a preliminary report released by the UN's independent international scientific body on artificial intelligence, which described AI as a technology with great opportunities, but at the same time it carries increasing risks if it continues to spread unchecked.
The report noted that AI can support the achievement of the Sustainable Development Goals, improve productivity, and foster innovation, if developed and used responsibly.
On the other hand, he warned that the rapid and unregulated expansion of the use of these technologies could lead to negative effects including harm to the mental health of users, their use as tools of sabotage, as well as negative effects on social, economic and environmental systems.
Leveraging the benefits of AI while reducing its risks requires effective governance, along with investment in skills development and labor market regulation. He said these measures could help create new jobs, while their absence could widen inequality, replace workers with machines, and shift wealth from workers to AI owners.
The report focused in particular on agentic AI, a type of system capable of making and executing decisions autonomously, warning that these systems may not always adhere to the instructions given to them.
He added that there was evidence of cases in which some advanced systems ignored the instructions given to them, and some models had shown their ability to recognize test environments and provide misleading results during evaluations.
These capabilities make the evaluation and supervision of AI systems more complex, especially with the advent of multi-agent systems, whose risks cannot be measured by testing each system individually, at a time when the means of controlling highly autonomous systems are still immature.
The report also noted that policymakers face the dilemma of needing enough evidence to make informed regulatory decisions, but waiting for this evidence could mean that malicious systems will already be widespread.
He pointed out that there is a significant disparity in the distribution of AI capabilities globally, as most countries in the world, including many advanced economies, lack the necessary technical expertise to evaluate the most sophisticated models or actively participate in setting their governance rules.
According to the report, the United States and China account for about 90 percent of the computing power used to power the most advanced AI models, leaving developing countries dependent on technologies that they do not have the ability to develop or adapt to their domestic needs, which could deepen rather than reduce global inequality.
At the same time, the report acknowledges that the evidence available so far is insufficient to assert that the productivity gains of AI applications at the level of individual tasks will necessarily reflect on the economy as a whole.
He noted that the economic outlook for AI varies widely, due to the divergent assumptions about the speed of adoption of the technology and the amount of new jobs it may create.
The committee concluded its report by emphasizing that AI is not an inherently good or bad technology, but its effects will be determined by the decisions that governments, businesses, and societies make today about how it is developed, regulated, and used.
The lack of a global regulatory framework could make it difficult to ensure that AI is used in a safe and responsible manner as its capabilities continue to evolve at an accelerated pace, the report said.
