The fAIrgov Project

Public Fairness Perceptions of Algorithmic Governance (fAIrgov)

Short description: Algorithms based on artificial intelligence are increasingly being used by governments to make decisions that impact citizens’ lives. While significant attention has been devoted to theoretical discussions on fairness, accountability, transparency and trust, less is known about citizens’ perspectives. The fAIrgov project has since 2018 tracked the views of citizens, poiticians, and public servants in Norway on matters such as knowledge about AI and support for using AI in the public sector.

Key preliminary findings are that knowledge of AI has increased over the years among citizens, and with it also support for using it. Significant concerns are however also expressed, and these concerns are unevenly distributed along party lines and socioeconomic demographics. Elected representatives are in general well aligned with their voter bases, with some markable exceptions.

Host institution: NORCE – The Norwegian Research Centre. Partner institutions: Stanford University University of Bergen

Funder: Research Council of Norway (project no. 314411)

Researchers: Sveinung Arnesen (Principal Investigator), Mikael Poul Johannesson, Troy Saghaug Broderstad, Henrik Litleré Bentsen, Jon Kåre Skiple, Gjøri Marie Haugen, Anne Marthe Borgen, James S. Fishkin, Alice Siu

Data generating infrastructures: Coordinated Online Panels for Research on Democracy and Governance (KODEM), Stanford Online Deliberation Platform

Research

Knowledge and support for AI in the public sector: a deliberative poll experiment

Abstract: We are on the verge of a revolution in public sector decision-making processes, where computers will take over many of the governance tasks previously assigned to human bureaucrats. Governance decisions based on algorithmic information processing are increasing in numbers and scope, contributing to decisions that impact the lives of individual citizens. While significant attention in the recent few years has been devoted to normative discussions on fairness, accountability, and transparency related to algorithmic decision-making based on artificial intelligence, less is known about citizens’ considered views on this issue. To put society in-the-loop, a Deliberative Poll was thus carried out on the topic of using artificial intelligence in the public sector, as a form of in-depth public consultation. The three use cases that were selected for deliberation were refugee reallocation, a welfare-to-work program, and parole. A key finding was that after having acquired more knowledge about the concrete use cases, participants were overall more supportive of using artificial intelligence in the decision processes. The event was set up with a pretest/post-test control group experimental design, and as such, the results offer experimental evidence to extant observational studies showing positive associations between knowledge and support for using artificial intelligence.

Suggested citation: Arnesen, S., Broderstad, T.S., Fishkin, J.S. et al. Knowledge and support for AI in the public sector: a deliberative poll experiment. AI & Soc (2024). https://doi.org/10.1007/s00146-024-02104-w

Differences between Citizens, Elected Representatives and Public Administrators Attitudes Towards AI in the Public Sector

Abstract: This study compares attitudes toward Artificial Intelligence (AI) use in the public sector among elected representatives, public administrators, and citizens. Elected representatives enact policies that facilitate AI integration in decision-making, while public administrators implement these policies, often with greater insight into how AI affects their roles. Citizens, on the other hand, are directly impacted by these decisions and hold elected officials accountable for their AI governance. Given the increasing use of AI in public administration, understanding how attitudes differ across these groups is important: Misalignment could lead to overstepping in algorithmic governance that may erode trust in public institutions over time. We surveyed a representative panel of citizens (N=4206), elected representatives (N=1898), and public administrators (N=2979) in Norway. We assessed their knowledge about AI, whether they believe AI will improve or worsen public services, and their support for using AI in resettling refugees. Our findings show public administrators are most optimistic about AI improving public services, have the highest self-reported knowledge, and are most supportive of using AI in refugee resettlement. Elected representatives are more positive than citizens but less so than public administrators. Citizens are the most sceptical of, reporting the lowest levels of AI knowledge. These results highlight significant differences and have important implications for algorithmic governance.

Keywords: Artificial intelligence, Congruence, Bureaucrats

Suggested Citation: Broderstad, Troy S. and Arnesen, Sveinung and Johannesson, Mikael Poul, Differences between Citizens, Elected Representatives and Public Administrators Attitudes Towards AI in the Public Sector (February 06, 2025). Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5126671

Who Deliberate? Participation and Recruitment in Deliberative Mini-Publics

What characterize citizens who choose to participate in deliberative mini-publics? Using population registry data from Norway, we compare socio-demographic attributes among participants and non-participants in two Deliberative Polls conducted in 2021 and 2022, the latter on the topic of using AI in the public sector. We find that most citizens decline invitations to take part, whereas those who do participate are fairly representative of the population even without targeted recruitment strategies such as using quotas or stratified sampling. Some social groups that have shown low participation in elections-such as younger cohorts and immigrants-are well represented in the deliberative mini-publics without needing any extra recruiting efforts, showing that it indeed is possible to combine the democratic principles of equal opportunities and representative deliberative mini-publics. Following up with conjoint experimental data, we find that increasing the monetary incentives increase willingness to participate, as do events where citizens are invited based on random selection rather than pure self-selection. Finally, willingness to participate increases when participants are not required to publicly reveal their attitudes on the topics under consideration. The study contributes both to the study of representation in deliberation as well as to practical knowledge about recruitment and participation in deliberative mini-publics.

Keywords: Deliberative mini-publics, Representativeness, Participation, Recruitment, Conjoint analysis, Population registry data

Suggested Citation: Arnesen, Sveinung and Skiple, Jon Kåre, Who Deliberate? Participation and Recruitment in Deliberative Mini-Publics (February 10, 2025). Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5130755