UN’s ambitious agenda for AI’s future landscape

As the United Nations General Assembly convened in New York this week, Amandeep Gill, the UN Secretary-General’s representative on technology, presided over an occasion named “Governing AI for Humanity.” During this event, attendees engaged in discussions regarding the potential perils associated with AI and the complexities of attaining global collaboration in the realm of artificial intelligence.

Secretary-General António Guterres and Gill have expressed their belief that the establishment of a new UN agency will be necessary to facilitate global collaboration in overseeing this influential technology. However, the specific concerns the prospective entity would aim to tackle, along with its framework, remain undetermined. Moreover, certain analysts suggest that ambitious initiatives for worldwide cooperation, such as this, seldom receive the necessary backing from influential countries.

UN’s role in fostering collaboration

Gill has previously spearheaded endeavors to enhance the safety of sophisticated technologies. During his tenure as chair of the Group of Governmental Experts of the Convention on Certain Conventional Weapons, the Campaign to Stop Killer Robots, aimed at urging governments to ban the production of lethal autonomous weapons systems, encountered obstacles in garnering support from major global powers like the US and Russia. Currently, Gill is guiding a more ambitious initiative to promote global collaboration in the realm of artificial intelligence.

The advancement of AI has accelerated significantly in recent years, and industry experts anticipate this rapid growth to continue unabated. The far-reaching effects of AI are projected to transcend national boundaries, prompting global leaders and technology experts to advocate for international collaboration concerning AI-related matters.

During a UN Security Council assembly in July, Guterres advocated for the United Nations as the suitable platform for fostering this collaboration. The subsequent progression towards the creation of a UN AI agency involves the formation of the High-Level Advisory Body on Artificial Intelligence, the constituents of which will be disclosed in October.

Having previously acted as the executive director of the UN High-Level Panel on Digital Cooperation between 2018 and 2019, Gill assumed the role of tech envoy in June 2022. In August of the present year, the tech envoy’s office initiated the recruitment procedure for specialists to join a High-Level Advisory Body on Artificial Intelligence, inviting the public to submit nominations.

Also Read: Can you bypass ChatGPT restrictions and get an unfiltered AI chatbot?

Debate over new AI agency

TIME viewed a distinct request for nominations that was dispatched to member states, detailing the guidelines for the formation of the Body. These guidelines stipulated that the Body would comprise a maximum of 32 individuals from various sectors, encompassing diverse gender representations, age groups, geographic locations, and areas of expertise.

In an interview with TIME on August 30, Gill disclosed that his office had garnered over 1,600 submissions via the public call for nominations. Anticipating an aggregate of over 2,000 nominations, including those from member states, he mentioned that the tech envoy’s office, in collaboration with other UN entities, will compile a list of finalists, from which the Secretary-General will choose the final 32 members. The inaugural gathering of the Body is slated for October.

The guidelines outlined in the document reviewed by TIME and validated by Gill indicate that the Body is tasked with preparing an initial report, presenting a comprehensive evaluation of alternatives for the global regulation of artificial intelligence by December 31, 2023. A subsequent report, due by August 31, 2024, is expected to offer comprehensive recommendations concerning the roles, structure, and proposed timetable for a prospective international agency governing artificial intelligence.

Aki Enkenberg, heading the innovation and digital cooperation team at Finland’s Ministry for Foreign Affairs, characterized the stipulation directing the Body to propose a new international agency as a “rushed decision.” 

He contends that a significant portion of the AI-related global governance could be effectively managed by existing entities within the UN system, stressing the importance of conducting an analysis to identify any potential gaps before advocating for the establishment of a new agency. Gill refuted the notion that the terms of reference would prejudice the Body’s conclusions in favor of recommending the formation of a new agency.

In response to concerns regarding the potential unwieldiness of the advisory body’s size, which might inadvertently grant the secretariat excessive sway over the report’s content, Gill emphasized that the secretariat has no intention of exerting influence over the conclusions drawn by the Body.

Also Read: EU’s AI Act might be bad news for small firms

Exploring AI governance models

By September 2024, during the UN’s Summit of the Future, Gill anticipates that the conclusions drawn by the Body will equip member states with the necessary insights to determine whether and how to endorse the creation of a UN AI agency.

In a May blog entry, executives of OpenAI, the organization responsible for ChatGPT, contended that the future might necessitate an entity akin to the International Atomic Energy Agency (IAEA) for the purpose of effectively regulating increasingly complex artificial intelligence systems.

Established in 1957, the International Atomic Energy Agency (IAEA) conducts inspections aimed at preventing non-nuclear weapon states from acquiring nuclear weapons, while also offering technical assistance for the peaceful advancement of nuclear energy.

However, emulating the IAEA’s model for artificial intelligence is merely one prospective avenue. Referencing a paper during the interview with TIME, Gill highlighted that in July, researchers from renowned AI labs such as Google DeepMind and OpenAI, along with several academic and nonprofit research institutions, presented four distinct potential structures for an international AI entity, all of which are not mutually exclusive.

Challenges in establishing AI oversight

One possible proposition is the establishment of an agency similar to the IAEA, capable of formulating industry-wide benchmarks and overseeing stakeholders to evaluate their compliance. Another proposal entails creating an entity fashioned after the Intergovernmental Panel on Climate Change (IPCC), fostering expert consensus on technical matters. 

The report also recommends an international coalition between the public and private sectors, guaranteeing equitable access to beneficial AI, akin to Gavi, the Vaccine Alliance’s approach to vaccines. Lastly, the suggestion involves initiating an international collaborative effort for AI safety research, reminiscent of the European Organization for Nuclear Research (CERN).

Certain experts contend that garnering support for an IAEA-style agency for AI among policymakers would be challenging. This is due to the fact that it would necessitate countries like the US and China to grant international inspectors unrestricted entry to their most cutting-edge AI laboratories, all in an attempt to mitigate risks that have yet to materialize.

Containing AI safety risks within national boundaries might prove challenging, underscoring the necessity for the inclusion of all countries involved in the development of potent AI systems, particularly the US and China, in any global accord. Sihao Huang, a University of Oxford researcher who delved into AI governance in Beijing over the past year, suggests that a concerted emphasis on common concerns, such as “assessing models for biosecurity risks,” will be crucial for securing China’s support for an international agreement.