Technological innovation has historically emerged from concentrated centers of expertise, developed by a limited group of specialists with minimal public oversight or legal frameworks (e.g. printing press, nuclear weapons). The development of artificial intelligence—defined here as machine learning models trained on data that can be deployed for real-world applications, encompassing both supervised and unsupervised approaches, whether predictive or generative—follows this pattern, with development concentrated among a small number of private companies and research institutions [1].

However, AI systems are increasingly deployed in domains that affect fundamental aspects of human life, from healthcare and education to employment and criminal justice [2, 3, 4]. This broad societal impact, combined with AI's potential to reshape power structures and decision-making processes, raises critical questions about its governance. A stakeholder-based analysis suggests that those affected by AI systems should have a voice in shaping both their development and deployment.

Yet this proposition faces significant challenges. The technical complexity of AI systems creates a substantial knowledge gap between experts and the general public. This disconnect raises fundamental questions about the feasibility of democratic participation in AI governance: How can meaningful public input be incorporated into decisions that require deep technical understanding? What mechanisms could bridge the divide between scientific expertise and democratic legitimacy?

Democratic governance, defined here as decision-making processes that ensure representation, accountability, and participation from affected stakeholders, presents both opportunities and challenges in the context of AI. This analysis will examine competing schools of thought: first exploring arguments for democratic governance, then considering the case for expert-led oversight. Ultimately, we will interrogate the possibility of integrating direct democratic participation into AI governance frameworks. Drawing on case studies from France and Brazil, we propose recommendations for a balanced governance model tailored to European Union countries.

Key words: AI governance, expert-led governance, participatory democracy, deliberative democracy

1. The Case for Democratic AI Governance: A Stakeholder-Based Analysis

1. A) Unive

...

Posts

Wikitag Contributions

Comments