A Bourdieusian theory on communicating an opinion about AI governance

This paper examines an often overlooked yet significant threat to survey validity and epistemic justice: the unequal com- munication of opinion. We discuss research that signals the presence of this threat when studying public opinion about AI. Furthermore, we apply Bourdieu’s theoretical framework as a potential explanation of the inequality in communicating an opinion about AI. We describe this inequality and test our explanation by performing a multilevel analysis on four questions about AI governance from the Eurobarometer 92.3 and two questions on its implications on our way of life and jobs from the Eurobarometer 95.2. Our results suggest that there is inequality in communicating opinions: higher social positions are more likely to communicate an opinion. We also find evidence to support the claim that the habitus is the underlying mechanism mediating this inequality in opinion Our results suggest significant effects of self-perceived social class, external political efficacy, internal political and scientific efficacy, and relevant cultural capital regarding science and technology. Lastly, we do not find consistent results regarding the effect of the selected contextual level variables across the two surveys. This paper examines an often overlooked yet significant threat to survey validity and epistemic justice: the unequal communication of opinion. We discuss research that signals the presence of this threat when studying public opinion about AI. Furthermore, we apply Bourdieu’s theoretical framework as a potential explanation of the inequality in communicating an opinion about AI. We describe this inequality and test our explanation by performing a multilevel analysis on four questions about AI governance from the Eurobarometer 92.3 and two questions on its implications on our way of life and jobs from the Euro- barometer 95.2. Our results suggest that there is inequality in communicating opinions: higher social positions are more likely to communicate an opinion. We also find evidence to support the claim that the habitus is the underlying mechanism mediating this inequality in opinion our results suggest significant effects of self-perceived social class, external political efficacy, internal political and scientific efficacy, and relevant cultural capital regarding science and technology. Lastly, we do not find consistent results regarding the effect of the selected contextual level variables across the two surveys. Our findings suggest that inequality in communicating an opinion is widely present when studying public opinion about AI. Future stud- ies should check for this inequality before widely distributing their surveys. Should such inequality be detected, corrective measures should be taken to preserve research validity and mitigate epistemic injustice. Our findings suggest that inequality in communicating an opinion is widely present when studying public opinion about AI. Future studies should check for this inequality before widely distributing their surveys. Should such inequality be detected, corrective measures should be taken to preserve research validity and mitigate epistemic injustice.