Rethinking Policy Frames of AI for Societal Challenges: from Technical Fixes to Co-Production

In policy debates about risks and opportunities associated with Artificial Intelligence (AI), one powerful frame emphasizes the potential of AI to tackle major societal challenges of our times in areas such as health, environment, and energy. While this is a hopeful frame, it is far from straightforward what counts as societal challenge and what kind of AI policy would help to address them. In recent policy, societal challenges are often understood as the 17 United Nations Sustainable Development Goals which include important issues such as good health and wellbeing, quality education, and gender equality. But is this a comprehensive list of societal challenges of today? And how are these challenges understood across different regions, political systems, and cultural contexts? Moreover, AI is often presented as a quick technical fix to highly complex and uncertain societal problems, also known as ‘wicked problems’. This paper invites to rethink the policy framing of AI for societal challenges in more democratic and inclusive ways as a co-production involving a broad range of diverse social groups and affected communities.