SKILLNADER I DEN UPPLEVDA FÖRMÅGAN ATT IDENTIFIERA AI-DESINFORMATION. En kvantitativ analys som undersöker individens egen uppfattning om sin förmåga att upptäcka AI-genererad desinformation, samt hur denna uppfattning skiljer sig mellan människor med olika socioekonomiska förutsättningar

Sammanfattning

The swift development of artificial intelligence (AI) has fundamentally changed today’s media environment. Content generated by AI is increasingly present in digital platforms and news flows, making the distinction between authentic and manipulated information way more difficult for the audience. As a result, disinformation generated by AI has emerged as a growing societal challenge, with potential consequences for democratic processes, informed decision-making and trust in news media. Against this background, this study aims to examine how individuals perceive their own ability to identify disinformation generated by AI and how this self-perceived ability varies between different socioeconomic groups in Sweden. The main purpose of the study is to analyze differences in self-perceived ability to identify AI-generated disinformation based on income, educational level and place of residence. The study is based on theories of news literacy, which emphasize individuals’ active role in evaluating and interpreting news content, as well as the concept of self-efficacy, which refers to individuals’ confidence in their own ability to manage certain tasks. While previous research has shown that AI-related and digital qualifications are unevenly distributed across socioeconomic groups, considerably less attention has been paid to how people perceive their own ability to discover disinformation generated by AI. This gap in literature as well as previous research constitutes the central motivation for this study. Methodologically, the study is based on a quantitative research design based on secondary data from the Swedish National SOM Survey completed in 2023. The data consists of a representative selection of the Swedish population aged between 16-90 years old. The dependent variable of this study measures respondents' self-preserved ability to distinguish desinformation generated by AI. The independent variables include income, education level and residential region, while gender and age are included as control variables. The analysis reconciles descriptive statistics with ordinary least squares (OLS) regression models to identify and examine group differences. The results show that the overall self-perceived ability to identify disinformation generated by AI is proportionately low. A majority of the respondents placed themselves in the lower categories of the scale or in the middle, indicating extensive uncertainty in regards to their own ability. In terms of socioeconomic variables, educational level is the only factor showing a statistically significant conjunction with self-perceived competence. Individuals with lower levels of education tend to rate their ability to identify disinformation generated by AI more negatively than those with higher levels of education, even when verifying for other factors. In contrast, no significant connections are found either between income or place of residence and self-perceived ability. The control variables show that women tend to rate their ability lower than men, while younger individuals declare higher confidence than older respondents. This result suggests that self-perceived digital capability is shaped not only by structural means but also by external and psychological factors. Overall, the results indicate that distinction in self-perceived competence to identify disinformation generated by AI are primarily associated with educational level rather than place of residence or income in the Swedish context. This highlights the significance of educational initiatives focusing on strengthening critical information and media literacy in a media landscape increasingly driven by AI. By focusing on self-perceived capacity rather than actual ability, the study contributes a complementary perspective to existing research and increases understanding of how individuals navigate the complex environment of digital information.