A central AI secretariat, a standing scientific advisory board for artificial intelligence, two annual governance dialogues and other processes relating to the current hot topic – are the ideas of a high-level working group set up by UN Secretary-General António Guterres. At the upcoming UN Future Summit, the UN wants to get the green light from the member states for a new role. On Thursday night, however, negotiations on the Global Digital Compact (GDC), which is to include the AI recommendations, were still ongoing.
Anzeige
By UN standards, the working group has made the impact the UN Secretary-General had hoped for at the speed of light. The 39-member panel appointed by Guterres, which included AI company representatives from OpenAI, Microsoft, Google and Alphabet, AI experts from the Mexican Constitutional Court, the Vatican and renowned universities, as well as many government representatives, had around ten months.
Global raw materials, global use
The experts came from 33 countries. An attempt by the UN to counteract the problem of the lack of representation of countries from the Global South in existing AI governance approaches.
118 of the 194 UN member states are not yet involved in any of the regulatory instruments developed in the industrialized nations and AI pioneer states. The UN also wants to solve this representation problem with its new role. After all, according to the report, both the material from which AI is made, from rare earths to data, and the use of AI are global.
A graphic on AI regulations.
(Image: UN)
AI "climate" council, AI governance dialogs, AI fund
Just in time for the Future Summit, they presented their report with a total of seven institutional recommendations.
Anzeige
In future, an AI Council based on the model of the IPCC is to present annual reports on technological developments, research results and shortcomings. The new body is also to produce quarterly reports on topics relating to AI applications for achieving the UN development goals. Newly emerging risks will also be addressed in ad hoc reports.
The aim is to remain agile and provide a scientific basis for any governance decisions, said Carme Artigas, former Spanish Minister of State for Digitalization and AI, at the press conference in New York on Thursday evening. It is also not about replacing other institutions or national regulatory authorities, but about aligning and converging their practices.
Green light for proposals?
The AI Council has incorporated it directly into the draft resolution for the Global Digital Compact (GDC), the new United Nations Digital Charter. The member states are to decide on this at the Future Summit. According to the GDC draft, the members are to be nominated by member states in consultation with stakeholders.
The GDC has also taken up the idea of biannual AI governance dialogs to exchange best practice and abuse scenarios and attacks. The same applies to the idea of a fund to be set up with state and private resources to promote work in the UN and the development of knowledge and capacities worldwide.
Proposals not (yet) adopted
The four other recommendations made in the expert report were not explicitly included in the most recently published GDC draft.
An AI office reporting directly to the UN Secretary-General, for example, is probably going too far for some member states. Instead, the members of the High-Level Panel saw it as a central hub to support the various initiatives, from the AI Council to capacity building. The Capacity Building Network, which is intended to provide start-up support for AI newcomer and latecomer countries, and the idea of developing a data governance framework for AI are missing. However, a separate chapter in the GDC is dedicated to the topic of data exchange and data flows.
The expert panel could imagine more in terms of technical standardization. There is no mention of the UN as a kind of clearing house for AI specifications (most standards have so far been drafted by the ITU) in the most recently published GDC version. Instead, the AI organizations themselves are called upon to coordinate more.
No dedicated UN AI authority
The original idea of giving the UN its own fully-fledged AI authority in the style of the UN Nuclear Regulatory Authority has been shelved for the time being. The UN itself had toyed with the idea. "For now", the panel has decided that the UN does not need such an agency, said Indian diplomat Amandeep Singh Gill, Tech Envoy of the UN Secretary-General, at the press conference.
However, the new AI initiatives, which are to be adopted early next week, will hopefully lay the foundation for the UN's role in the field of AI. "In the future, when the risks associated with AI make certain functions necessary that can be taken over by such an institution, this can be reconsidered," says Gill.
