The opportunities and risks of Artificial Intelligence (AI) seem limitless. While some believe that many of the most challenging problems of this world can be solved through or with the help of AI, other, such as Bill Gates, consider this technology also a risk alike to atomic power and atomic weapons.
There are a number of reasons why AI poses a global risk: First, AI is analysing data through Algorithms, which are supposed to learn, and thus improve their performance, beyond capabilities of humans. While today such performance of AI beyond capabilities of humans is normally limited to very specific tasks, there is a trend to the broadening of these tasks. It cannot be excluded anymore that a general AI is being developed, which surpasses all aspects of human intelligence and eventually thus could dominate humankind in all respects. The problem of control of AI in order to ensure that humans do not become objects of machine control and the ability of states to govern is not undermined has already been amply described in science. It is clear that technology alone cannot deliver such control to a sufficient extent.
Second, much of Artificial Intelligence is and will be delivered and deployed via the Internet, across the borders of this world. It is thus a technology crossing borders in the virtual space without effective controls.
Third, AI is a technology, which is being developed by global corporations and states for various purposes and not limited to a sector. It is a multi-purpose technology, which has potential to scale globally in most areas of our lives, ranging from education and health via the production and delivery of media content and opinions, important for democracy, right through to the management of all essential infrastructures and military purposes. It is the sensitivity of the multiple contexts within which AI is deployed which requires to give its good functioning and governance highest attention. It is also clear that those who control the functioning of AI in sensitive sectors will command greatest power in this world.
Fourth, AI can be embedded in autonomous machines, which may cross borders in trade, crime or military operations, thus combining classical physical safety and security risks with the new risks of AI.
It is important to consider this cross border, global and multipurpose nature of AI in any attempt to assess risks and opportunities of AI. The world has become a community of risk, not only relating to COVID, but also relating to AI. And the scarcest resources is not the one next great idea in terms of a technological solution to a major global challenge, but the scarcest resource relating to AI is the ability to agree, both within states and among states, on how to govern this new powerful and globally scaled multipurpose technology to the benefit of states and mankind.
It is before this background that we are discussing how to give an impetus to the international community and the United Nations to start work on a global multilateral agreement on the governance of AI.
Paul Nemitz, Distinguished Contributor to the book Remaking the World – Toward an Age of Global Enlightenment, will be the lead speaker of Plenary I at the Club de Madrid – BGF Policy lab, September 7, 2021.
Please register here: https://aidigitalrights.com/registration/