To better understand what drives the public’s perception and acceptance of AI in different roles, we propose a study that looks at varying AI domains by occupational status and individual differences across ontological perceptions, automation anxiety, perceived status, and identity threat. As a first step, we conducted a representative survey of the US population (N = 1,005) that looked into the public's perceptions of AI replacement of high-status jobs. Results indicate that a majority of participants hold negative attitudes about AI replacement in all domains presented. However, participants were more open to AI replacement in lower-status roles such as journalist and hiring manager compared to higher-status roles of spiritual leader and trial judge. Contrary to our expectations, participants believed that trial judge was a slightly worse idea than AI spiritual advisor. This finding suggests that the associated machine heuristic of the judge role as being a more rational and objective occupation was not triggered in our sample. Our results also suggest that more vulnerable populations are more reluctant to accept AI in the majority of jobs. These findings are in line with previous public opinion surveys and demonstrate that individuals with lower levels of power and status are more likely to be reluctant to accept new technology and potentially perceive it as a threat. Our next step will be to include more occupations that can be potentially automated and look for explanatory mechanisms driving the public’s view of AI integration.

Internet users face privacy threats when using online services. Privacy protection behaviors, such as adjusting privacy settings, can alleviate some of these threats. Research shows that individuals’ privacy protection behaviors (PPBs) depend on their socio-demographics characteristics, digital engagement, privacy concerns, and online privacy literacy (OPL). In addition, it has been suggested that due to the complexity of privacy issues online, an adequate level of OPL is required to translate privacy concerns into protective actions. Although previous research examined the antecedents of PPBs at a general level, it has rarely made a clear distinction and comparison between PPBs aimed toward the practices of institutions (vertical level) and those aimed toward other internet users (horizontal level). This is somewhat surprising given that many scholars underscored the importance of context in online privacy-related matters. Therefore, this study compared the antecedents of PPBs at the general, vertical, and horizontal levels. To this end, we tested three models to examine how socio-demographic characteristics, digital engagement, privacy concerns, and OPL influence PPBs at the general, vertical, and horizontal levels, and assessed whether OPL moderates the relationship between privacy concerns and PPBs at different levels. The models were tested using linear regression on a nation-wide sample of 1,015 internet users aged 18+ from Slovenia. The analysis revealed important differences between the levels in case of gender, age, and privacy concerns, but not OPL.

The following paper engages with citizen attitudes towards automated decision-making (ADM) in the public sector. Based on three domain specific scenarios, we explore differences in attitudes towards algorithmic automation in three welfare regimes, namely Estonia, Germany, and Sweden. It is notoriously different for citizens exposed to algorithmic automation by public sector institutions to evaluate the implications of these technologies. First, only a minority of citizens is aware of automated decision-making in the public sector, second the lack of transparency in the implementation process makes it difficult for citizens to develop an informed position. In order to explore the welfare regime and domain specific attitudes of citizens, we have worked with three scenarios that provide a concrete and empirical entry point for the respondents. The three scenarios include the use of ADM to sort job seekers into different categories, the use of risk scoring for child welfare and the use of facial recognition for predictive policing. Here, we present preliminary findings of the comparative analysis presenting first descriptive findings from the cross-regime comparison of attitudes towards ADM in the three scenarios and second a regression analysis including individual variables (age, gender, education) combined with awareness, enthusiasm, and trust for ADM systems to explain differences between the three welfare regimes.

In this paper, I provide the most comprehensive longitudinal survey of social media and elections so far, spanning across 15 years and covering five Danish national elections. Denmark is used as a critical case study as the country has a high Internet penetration, widespread use of digital media in public administration and citizen services, and a traditional high willingness to employ digital media in daily life. Thus, Denmark forms an exemplary case for longer trends for media use in election campaigns.

The uniqueness of this study is that is based on five repetitive survey questionnaires among Danish citizens across a time span of 15 years. By replicating questions across elections, one gets a comprehensive and fully comparable analysis of changing uses and attitudes towards social media in election campaigns.

Besides giving a statistical overview of social media use and attitudes across five elections, the paper addresses the following research questions:

To what extent have social media been “normalized” as an integrated part of election campaigns, as one media type among many?

Can we identify changing patterns of social media participation, from “clicktivism” (Halupka, 2017) to more formalized political participation or vice versa?

How do social media, over time, contribute to political interest, motivation, and competence among citizens?

In sum, the paper provides a longitudinal study across five elections of voters’ social media use in a digitally advanced country.