Copilot skrev alla svaren : – en jämförande studie av lärobokstext och AI-genererade svarsom kunskapskälla i religionskunskap på gymnasiet

Sammanfattning

This thesis examines what kind of knowledge about religion is conveyed by twoknowledge sources in Swedish upper-secondary religious education: a current textbookand a generative AI tool. As pupils increasingly turn to AI to have study questionsanswered, sometimes instead of reading the textbook, the character of the knowledgethey encounter becomes important. The aim is to analyse and compare the knowledgeabout religion conveyed in two chapters of Religion för gymnasiet nivå 1 & 2 and inanswers generated by Microsoft 365 Copilot, and to examine how these representationscan be understood as powerful knowledge. The theoretical framework is MichaelYoung's theory of powerful knowledge, operationalised through three criteria: whetherknowledge moves beyond everyday experience, whether it is presented throughsystematically related subject concepts, and whether it reflects a specialised,disciplinary perspective. The method is a qualitative content analysis using the samestudy questions as a shared unit of analysis for both materials. The analysis shows thatboth sources cover largely the same content, since both depart from the same studyquestions, but differ in how it is represented: the textbook builds knowledge gradually,defines subject-specific concepts and situates phenomena in historical and societalcontexts, whereas Copilot's answers are more summary and often leave subject-specificconcepts implicit. The study concludes that the textbook to a greater degree fulfils allthree criteria of powerful knowledge, while Copilot's answers display features of it butmeet the criteria to a lesser extent, particularly regarding conceptual structure anddisciplinary grounding.