AI Governance: How an international framework can derive from business standardsand multilateralism

While there is no constraining international framework regarding the use and development of artificial intelligence, international institutions are starting to provide general guidelines destined both for governments and organizations. In this article, two of these frameworks will be studied: nominatively the Institute of Electrical and Electronics Engineers’ (IEEE) P7003 standard and the Global Partnership on Artificial Intelligence. (GPAI) Their main tenets will be compared. Afterwards, Singapore’s Model AI Governance Framework will be presented as a potential model for further development of AI regulation on an international scale.

IEEE P7003 is developed since 2017 as a global standard on “Algorithmic Bias Considerations.” Its mechanisms is that it provides developers of Ai processes with methodologies in order for their programs to be transparent and less biased as possible. There is no institutional enforcement as the IEEE cannot legally bind organizations. But the IEEE’s expertise is widely recognized in the industry, and IEEE P7003 is increasingly enforced through contracts, and is seen as an industry norm, reassessing its informal importance.

The GPAI, announced in 2020 by Macron and Trudeau in the context of a G7 meeting. It involves 29 countries, mainly Western, and the EU today. It does not provide a set of constraining rules, but offers a forum for collaboration in AI governance, and facilitates the sharing of knowledge in this domain, both between states and in public-private collaborations. Its goal is to advance the development of AI with reducing challenges that it can bring. In doing so they want to develop the commercial use of AI and facilitate its incorporation in daily life, while also emphasizing the importance of data governance and pushing for the development of global AI ethics, through working groups. Under the OECD, this program is open and inclusive to participation, and one if its main tenets is its evolution.

While the IEEE P7003 is opting for a business approach to formalize its standard as th global norm, the GPAI emphasizes on a multilateral approach to develop and legitimize their optics on the future of AI. Resting on countries’ interest in a realist approach to international relations, or on companiess economic interests can be beneficial to further the impmlementation of an international framework. But should we only rely on these tools, or should there be at least a part of state constraining ? The example of Singapore’s Model AI Governance Framework (MAIGF) will be used to answer this question and offer a synthesis.

With its first version implemented in 2019 and regularly updated, the MAIGF is developed by Singapore’s Personal Data Protection Comission at the destination of any organizations, to control their use of AI in a transparent way. Its main components include the establishment of a clear governance structure regarding AI in organization, the responsible use of data, and the preeminence of human decisions even when AI is used. While it is not technically binding, all organizations in Singapore are called to abide by these rules, so as to create a safe and trustworthy business environment. The national specificty of Singapore is that there are extremely clear and open business environment and regulations, which makes it in turn that complying to these rules is almost an imperative for any organizations wishing to partake in economic activities in the country, and by extension in the globalized economy.

While there is no constraining international framework on Ai global governance, it is interesting to note that many initatiatives are already on the way. All focused on issues of transparency and risk reduction in AI implementation, they rely on different coercive ways to implement their regulations. They all offer a partly effective way to regulate AI, but a mix of these three is the most effective solution, as it would combine private actors’ economic interests, a multilateral stance, and domestic regulations that have the power to influence the world.