Human-centric AI governance: what the EU public values, what it really, really values

The EU and other institutions worldwide have committed to aligning AI with human values to ensure that the technology contributes to the common good. Yet, criticism persists that debates over which values should guide this alignment are dominated by private and public organizations that prioritize technological considerations. Societal perspectives that emphasize broader, non-normative values are often marginalized. This exclusion generates a democratic deficit and risks forgoing the advantages of aligning AI with citizens' public values - namely trust, acceptance and public value creation. To address this gap, we empirically examine EU citizens' regulatory and value preferences regarding AI and its regulation, drawing on two complementary studies and Public Values theory and tools: one mixed-methods study of the EU's Public Consultation and one study based on the quantitative assessment of a newly developed AI Public Value (PV) Landscape. Our findings show that EU citizens (i) prefer binding regulation of AI, (ii) consider both ethical and technological principles as well as broader, non-normative societal values, especially along the moral-ethical value dimension, important, and (iii) serve as a conciliatory force capable of balancing business interests against those of state and NGO stakeholders. These results underscore the importance of aligning AI with broader PVs, reinforcing ethical foundations, and enhancing public inclusion in AI governance to achieve truly human-centric and socially accepted AI. • Public Value theory complementing AI ethics to achieve human-centric, accepted AI. • Newly developed AI Public Value Landscape to assess societal AI value preferences. • Citizens prefer bindingly regulated compared to unregulated AI. • Citizens appreciate extensive set of AI Public Values across all basic value dimensions. • Citizens as conciliatory force in AI governance and ethics debate.