Artificial Intelligence, Global Governance and Deliberative Democracy: Analysing, Evaluating, and Innovating Global AI Governance

Recent advances in artificial intelligence (AI) technologies have led to rapid development and deployment around the world. While AI promises improvements in areas like healthcare, education, transportation, and scientific discovery, it also poses significant global risks. These include rising inequality, technological unemployment, algorithmic bias, mass surveillance

and potentially long-term threats to safety. The transnational and dual-use nature of AI means national-level governance alone is insufficient for mitigating risks. As such, promoting legitimate and effective global governance of AI will be essential to mitigating the risks of the technology whilst promoting its benefits.

There are currently two analyses that comprehensively map out key actors and initiatives in the global governance of AI (Schmitt, 2022; Butcher and Beridze, 2019). Global AI governance tends to be characterised as polycentric and decentralised, with various actors and

initiatives attempting to influence norms and standards within the space. It is also largely topdown, with little to no opportunities for everyday citizens to participate in global AI governance (Schmitt, 2022). At the same time, there have been increasing explorations by scholars and practitioners on

how to 'democratise' the governance of AI (e.g., Seger et al., 2023). One approach to democratising AI governance is to deploy deliberative processes, such as citizens' assemblies (e.g., Landemore et al., 2023). This draws on the extensive normative and empirical research within the field of deliberative democracy and democratic innovation.

In practice, a range of deliberative and participatory processes have now attempted to engage citizens in AI governance, but not yet at the global level. This includes citizens' councils, participatory design processes and online crowdsourcing. In the coming months and years, this trend will continue with organisations like the Collective Intelligence Project, the Ada

Lovelace Institute, and the Iswe Foundation (amongst others) all planning to convene deliberative processes on AI governance. Despite these calls to democratise AI governance and the emergence of deliberative processes in this space, there has not been a detailed exploration of the role that deliberative democratic

innovations could play in addressing deficits in global AI governance. This dissertation seeks to address this in four steps. Below, I outline each step, the methodology for carrying it out, and why doing so is important for scholarship and practice.