Funding: Green Box, University St Gallen
Governments increasingly rely on algorithmic decision-making, with many AI systems developed and managed by private companies. While this creates various democratic concerns, we know little about how these governance choices affect citizen support for algorithmic decision-making (ADM). In this study, we investigate how different AI governance models—where AI is developed and managed by public actors, private companies, or co-developed with citizen input—affect public support for ADM. We theorize that governing actors serve as institutional signals, conveying expectations about value priorities embedded in AI systems and ADM. Three vignette experiments (N=4,784) in the US and UK manipulating AI governance (government vs. private vs. participatory) conducted between September 2024 and April 2025 in education and welfare contexts support this framework. Studies 1 and 3 reveal small but significant main effects with citizens preferring state-led over company-led AI development. In study 2, conducted following the US presidential transition amid prominent ties between technology executives and the new administration, treatment effects depend on citizen trust in government and private companies. These results demonstrate the importance of studying the institutional context in which AI is embedded, beyond technical specifications, for understanding public support for ADM. This research contributes to emerging literature on AI governance at a crucial moment when governments face fundamental choices about AI integration in public decision-making, institutional responsibilities for algorithmic oversight, and regulatory architecture
Funding body: Swiss National Science Foundation
Project Summary: Can democratic politics incorporate citizen demands for independent expertise in ways that boost legitimacy and trust in politics? Democratic governments worldwide face the dilemma of how to deal with an ever-increasing call for technocratic expertise needed to govern effectively while remaining responsive to and representative of the citizens who elected them. In the past decade, the economic crisis brought this tension to the forefront of democratic politics, with multiple appointments of technocratic ministers and governments across democratic states and a simultaneous populist backlash against an apparent “out-of-touch” political establishment. More recently, the climate crisis and the global COVID19 pandemic highlighted the role of independent scientific expertise in guiding political decisions, but also the crucial role that citizens’ attitudes play in shaping policy effectiveness and trust in politics. Despite the pressing and complex issues that governments need to address, it is uncertain how democratic politics can include more independent expertise in a way that increases public support for political processes and decisions. From the perspective of citizens, we are currently presented with the following empirical puzzle: on the one hand, citizen surveys show growing demands for independent experts over politicians in political decision-making and a recognition that complex global problems require experts to solve them. On the other hand, we observe mounting public skepticism towards technical knowledge and scientific expertise, paired with soaring anti-elite rhetoric stoked by populist actors across established democracies. How can we reconcile these conflicting observable phenomena and what solutions can we offer for reinforcing support for democratic politics? The ‘Varieties of Expertise’ project addresses this puzzle through three key research questions:
(I) What constitutes ‘politically legitimate’ use of expertise and who is considered an “independent expert” in the eyes of citizens?
(II) Why do citizens demand more political power in the hands of independent experts?
(III) How and where do citizens want to see political power in the hands of independent experts?
Overall, the ‘Varieties of Expertise’ project will contribute to efforts to ameliorate public responses to political decisions, decrease political polarization and re-build political trust between citizens and political actors across established democracies. Insights will be highly relevant for scholars and political practitioners concerned with the design of political arrangements that can boost public acceptance of policies in contested areas, such as health, environmental, economic and immigration politics.
More information: https://eribertsou.com/var-exp/
Talk at the Institute For Futures Studies
https://www.youtube.com/watch?v=S9myr0ElH2w
October 2023, Stockholm, Sweden
Project Summary: Our team is part of a global Many Labs project lead by Dr. Viktoria Cologna (Harvard University) and Dr. Niels G. Mede (University of Zurich).. This Many Labs study seeks to analyze the factors that affect trust in science, as well as the prevalence and correlates of science-related populist attitudes across countries. Specifically, we are interested in a) public opinion on the role of science in society and policymaking, b) the perceived goals and benefits of science, c) the ways individuals across the globe inform themselves and communicate about science, and d) climate change attitudes.
More information: https://projects.iq.harvard.edu/manylabstrustinscience/home
Project Summary: Public support for a democratic system of government is thought to be one of the main bulwarks against democratic backsliding. Yet much of what we know about support for democracy is based on survey questions about “democracy,” a term that varies in meaning across countries and likely prompts socially desirable responses. Instead, we propose a new approach to measuring support for democracy: using a battery of 17 survey questions, we ask respondents from 19 national samples to evaluate the more granular rights and institutions that collectively constitute liberal democracy. We find considerable heterogeneity across countries in how our items cohere, but any disjunctures typically reveal faultlines in political cultures that might be exploited by authoritarian actors. We further identify a core set of seven items that provide a reliable and valid measure of public support for liberal democracy across our different samples.
More information: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4622645
Project Summary: Losers’ consent is widely considered a key resource for the perceived legitimacy and stability of democratic political systems. However, recent concerns around democratic backsliding and elite attempts to undermine democratic processes demand a shift in focus to the responsibility of political winners: They are expected to recognize and reject procedural violations, even when they receive favourable outcomes. Existing studies suggest that citizens are quite permissive of undemocratic behavior as long as it serves their own preferences. But are there limits to what citizens, particularly political winners, will tolerate? In this paper we argue that “winners restraint” – the flipside of losers’ consent – is an essential resource in established and newer democracies. We test the existence and limits of winners restraint for accumulating procedural violations in the context of policy decisions. Using a survey and two experimental studies in the UK, we investigated how many violations citizens were willing to tolerate before dissenting. Respondents exhibit double standards, with support for the winning side as the main predictor of how individuals perceive the legitimacy of political decisions. However, a majority of these winners demonstrate ‘winners’ restraint’ by revising their perceptions of fairness and legitimacy when violations accumulate.
More information: https://eribertsou.com/wp-content/uploads/2023/10/winnersrestraint_final_.pdf
