At the upcoming United Nations Future Summit in New York, a Global Digital Compact is set to be endorsed, focused on narrowing the digital disparity both within and across nations. The agreement will also require the compliance of the UN’s 193 member states in promoting responsible data management and in “improving the global governance of Artificial Intelligence (AI) for the greater good of mankind.”
The compact’s goals largely revolve around extending the perks of digital technologies, inclusive of AI, to underprivileged countries and less affluent societal groups. In terms of AI, the emphasis is not just on the potential advantages but also on the inherent risks of this technology. According to a draft of the agreement, “There is an immediate need for a comprehensive evaluation and understanding of the predicted impact, possibilities and dangers that AI systems pose to sustainable development and the welfare and rights of individuals.”
The US, the EU and China are all pivoting towards a risk-oriented way of lawmakers’ regulating AI, despite the intense rivalry in the field preventing them from establishing a shared framework. Thorsten Jelinek, a specialist in international digital governance from Hertie School in Berlin and Taihe Institute in Beijing, is of the opinion that an innovative perspective is called for. Jelinek suggests that the predominant risk-based approach reliant on legislation will not save us from widespread, highly potent AI, terming such a notion as “utopian.”
Instead, Jelinek favours the creation of an ethical framework that views AI as a moral counterpart to humans. However, he proposes a practical solution in the interim, citing the Open Skies Treaty as a model. This treaty permitted parties to perform unarmed aerial inspections over their territories to provide reciprocal assurance against potential attacks.
Jelinek explained his proposal stating, “Both parties would keep their AI models confidential, but they would eventually reveal the capabilities of those models to the other side.” According to him, “Once the other party learns what an advanced frontier AI is capable of, it won’t come as a surprise. It’s a simplistic approach, but certainly better than nothing.”