The United Nations has proposed establishing a global framework to monitor and regulate artificial intelligence (AI), treating it with the same urgency as climate change. This comes in a report released by the UN secretary general’s High Level Advisory Body on AI. The report recommends the creation of an AI monitoring body akin to the Intergovernmental Panel on Climate Change to gather information and assess the risks associated with AI.
The initiative aims to foster a new policy dialog among the UN’s 193 member states, focusing on identifying and addressing risks while exploring the potential benefits of AI. Special emphasis is placed on empowering nations in the global south to both contribute to AI governance and benefit from AI technologies. The recommendations include the establishment of AI standards, data-sharing systems, and training resources to assist these nations in AI governance.
The Global Digital Compact, an existing UN initiative designed to address digital and data inequalities, could facilitate some of these recommendations. Additionally, creating an AI office within the UN dedicated to coordinating these efforts is being considered.
According to Alondra Nelson, a professor at the Institute for Advanced Study and member of the UN advisory body, the international community largely agrees on the dual nature of AI’s potential—posing both opportunities and risks. The rapid advancements of large language models like ChatGPT have heightened concerns over AI’s potential to become uncontrollable, prompting over a thousand scientists and entrepreneurs to sign a letter calling for a six-month pause on AI development to assess these risks.
There are immediate worries concerning AI’s potential to automate disinformation, create deepfake content, replace human workers, and exacerbate algorithmic bias. The United States and China have already introduced competing resolutions at the UN for the development of “safe, secure, and trustworthy AI” and cooperative AI development respectively, both of which were signed by all member states.
As Joshua Meltzer from the Brookings Institute points out, given the competitive nature of US-China relations and differing priorities, universal agreement on AI governance may be limited. Additionally, AI technology is evolving too quickly for the UN to manage global cooperation singlehandedly, implying the need for a distributed architecture involving individual nations’ efforts as well.
Efforts in the EU to introduce sweeping AI regulations have already caused market disruptions, illustrating the divergent approaches among wealthy nations. California, reacting to the US’s hands-off approach, has proposed its own AI regulations, initially met with resistance from AI companies.
The UN report seeks common ground by emphasizing the alignment of AI governance with human rights, providing a strong basis in international law. Chris Russell, a professor at Oxford University, notes that the UN’s involvement could mitigate the redundancy currently seen in separate government evaluations of AI models.
While geopolitical competition is a factor, many scientists are unified in their call for increased cooperation on AI safety. This sentiment was echoed by prominent academics from the West and China in a recent joint statement issued in Vienna.
Ultimately, the success of this initiative will depend on the implementation of the UN’s recommendations. Nelson notes that while the blueprint for cooperation is promising, “the devil will be in the details of implementation.”
This article relies on information originally published by Wired.