Special Articles

Re-imagining AI policy for social benefit: three opportunities

Inga Ulnicane

Abstract

Development and implementation of AI policy around the world is taking place in the context of continuous hype characterized by high positive and negative expectations. While many of these expectations are likely later to turn out to be exaggerated, they influence decisions, actions and investments made today. Over the past five years AI governance has evolved from focus on the launch of voluntary ethics guidelines to the development and adoption of binding AI regulation. At international level, AI policy is developed in a number of existing and newly established interconnected forums. However, these forums are typically dominated by a small number of most developed and predominantly Western countries, while large parts of the world (mainly from the Global South) are not included. Against this background, this article outlines three opportunities for reimagining AI in more inclusive and socially beneficial ways. First, while AI today is characterized by huge power asymmetries, governments can undertake various roles to reshape power distribution in AI policy towards more equal relationships between government, industry and civil society. Second, an inclusive and democratic discussion of the purpose of AI development and use could help to shape it according to society’s values, needs and expectations. Third, the key social problems of our times such as climate change and inequality can be seen as ‘wicked problems’ that require holistic socio-technical approaches rather than just AI as a technological fix.