AI Suffrage: A four-country survey on the acceptance of an automated voting system

Abstract

Governments have begun to employ technological systems that use massive amounts of data and artificial intelligence (AI) in the domains of law enforcement, public health, or social welfare. In some areas, shifts in public opinion increasingly favor technology-aided public decision-making. This development presents an opportunity to explore novel approaches to how technology could be used to reinvigorate democratic governance and how the public perceives such changes. The study therefore posits a hypothetical AI voting system that mediates political decision-making between citizens and the state. We conducted a four-country online survey (N=6043) in Greece, Singapore, Switzerland, and the US to find out what factors affect the public's acceptance of such a system. The data show that Singaporeans are most likely and Greeks least likely to accept the system. Considerations of the technology's utility have a large effect on acceptance rates across cultures whereas attitudes towards political norms and political performance have partial effects.

Civicbase: An open‐source platform for deploying Quadratic Voting for Survey Research

“…The long-term vision of Civicbase is a place where people can create communities for eliciting preferences using novel methods like QVSR, this could take the form of municipal dashboards, citizen portals, worker collectives, and domain hobbyists. Civicbase could also be particularly helpful for the development of personal AI projects focused on political representation (Suter et al 2022). It will allow surveyors to improve the way they collect and aggregate preferences, and perhaps improve the accuracy and sensitivity of voting behavior predictive models.…”

“…Civicbase could also be particularly helpful for the development of personal AI projects focused on political representation (Suter et al. 2022). It will allow surveyors to improve the way they collect and aggregate preferences, and perhaps improve the accuracy and sensitivity of voting behavior predictive models.…”