Uppfattad kognitiv auktoritet hos generativ AI. Hur syntetisk information omvandlas till fakta

Sammanfattning

In this master’s thesis I aim to produce knowledge concerning the perceived expertise and cognitive authority of generative AI models, specifically in regard to health, politics and the environment. This problem is important to investigate in order to better understand how people interact with synthetic information, both as a reliable source and as an example of mis- or disinformation. Furthermore, it is important for both library and information science as well as the library profession to understand how generative AI changes the information behavior as well as the kinds of trust that people have towards (un)trustworthy and non-traditional sources. I use Patrick Wilson’s (1983) theory of cognitive authority and add to it an understanding of trustworthiness and competence by introducing theories of language and knowledge to better create a tool to investigate human-generative AI-interactions. I have conducted nine interviews and collected some of the interviewees’ prompts and answers with generative AI and used a combination of the theory of cognitive authority, thematic and textual analysis in my study. The results of the master’s thesis suggest that generative AI might be perceived as a cognitive authority in what people deem trivial matters, although there also seems to be great discomfort of treating generative AI as an expert or even human.