Användare i Generation Z och AI-genererat innehåll : Ett dual process-perspektiv på hur användare av sociala medier beslutsfattar vid identifiering av AI-genererade videor

Sammanfattning

The development of artificial intelligence (AI) in recent years has resulted in an increased amount of AI-generated content on social media. AI-generated content such as videos has throughout the development become increasingly realistic and more difficult to distinguish from authentic videos created by people. Due to this development, there is greater demand put on users' own ability to distinguish between AI-genererated and authentic videos. Prior research has primarily focused on a quantitative approach to explore how users identify AI-generated content such as images, audio or text, and not users’ thought processes. This study bridges the research gap by using a qualitative approach, to examine the decision-making of social media users when they identify AI-generated videos, to gain insight into which cognitive mechanisms affect decision making. A better understanding of this can be used to combat deception. Dual Process Theory, which proposes that human decision-making occurs either through type 1 (intuition) or type 2 (reflection), is used as a theoretical framework. Semi-structured interviews and think-aloud were conducted with eight respondents for data collection and resulted in a wide range of factors that the respondents used to identify AI-generated videos, as well as an insight into their decision-making processes. Thematic analysis of the empirical data resulted in three themes: human factors and intuition, technical aspects as confirmation of initial assessment and reduced confidence in authentic videos. The results of the study lead to the conclusion that users have a strong intuitive ability to read human attributes. During longer reasoning the users use technical details to confirm or contradict their initial assessment. Furthermore, the study showed that users who are insecure in their initial assessment risk overanalyzing and notice details that they perceive as revealing that a video would be AI-generated. This can result in the users assessing an authentic video as AI-generated. The study contributes with a new qualitative perspective on user decision-making when identifying AI-genererated videos.