Title: Who is that GAI?  – A qualitative study on Swedish journalists’ imaginaries of how generative artificial intelligence affects the professionalization of journalism.  The increasing discourse surrounding the role of generative artificial intelligence (GAI) in journalism has recently been widely discussed as it brings both opportunities and challenges for the journalism profession. How do technological advancements impact and challenge the journalism profession? This study aims to explore the imaginaries that Swedish reporters, AI-strategists and newsroom managers have regarding how the journalistic profession can change based on the new conditions and challenges that GAI has contributed to. Conducting a study on the subject based on a Swedish context is relevant to understanding Swedish journalists’ imaginaries of GAI and how an implementation of GAI may lead to a reconfiguration of previously given professional roles. The study answers to the following research questions:  What imaginaries do journalists in Swedish newsrooms have about generative artificial intelligence? How do the imaginaries about how generative artificial intelligence can affect the professionalization of journalism differ between reporters and managers or AI-strategists in Swedish newsrooms? Based on the interpretive nature of the research subject the study adopts a qualitative approach, hence the material was collected through eleven qualitative interviews. The empirical data was analyzed using the constant comparative method together with the study's theoretical framework consisting of sociotechnical imaginaries, professionalization and boundary work. The findings indicate that the Swedish journalists predominantly perceived GAI as a tool to enhance work efficiency, which results in the journalists being able to focus more on the creative parts of the profession. Furthermore, the journalists imagined GAI as a smart colleague to consult with, and the majority imagined that GAI primarily will function as a supplement rather than a replacement, and that GAI to some extent will reshape the profession.

The purpose of the study was to examine whether it was possible to discern any effect on the perception of the content of a copytext in terms of credibility and conviction, depending on whether the writer was ChatGPT or a human copywriter, and whether it was possible to discern any effect on the perception of the writer in terms of creativity and professionalism. To achieve the purpose of the study, the following questions were formulated: 1. How credible is an AI-generated text perceived compared to a text written by a human copywriter? 2. How convincing is an AI-generated text compared to a text written by a human copywriter? 3. How credible is ChatGPT compared to a human copywriter? 4. Is there a connection between the attitude towards AI and the perception of the content? The theoretical framework of the study consisted of theory and previous research regarding generative artificial intelligence (GAI), recommendations for writing a prompt, attitudes towards AI, the copywriting competence and recommendations for writing an effective copytext. The theoretical concepts of credibility and persuasion are two central concepts of the study that have been operationalised into competencies and recommended content as well as unique selling proposition (USP), call to action (CTA), rhetoric and storytelling. The results show that those who read the AI-generated copy text had a more positive perception of the content and the writer than those who read the human-written copy text. However, the results show that the creativity of the writer was perceived equally by both experimental groups. Overall, there were no statistically significant relationships between the perception of the content and the attitude towards AI. Based on the significance of the results and their mean values, the study's conclusions are that the text written by ChatGPT is perceived to be more credible in terms of competence and recommended content than the text written by the human copywriter and that the text written by ChatGPT is perceived to be more convincing in terms of USP, CTA, rhetoric and storytelling than the text written by the human copywriter. We can also conclude that ChatGPT is perceived to be more credible in terms of competence operationalised to professionalism, than the text written by the human copywriter, which, however, could be due to chance as the result was not statistically significant. The result of the study shows that ChatGPT threatens copywriting skills. However, it is not the AI tool itself that threaten the professional role, as it currently still requires human involvement. Instead, it is the competence in prompt engineering that threatens the profession, as this knowledge, according to the study, is crucial for achieving a credible and convincing AI-generated copytext. / Syftet med studien var att undersöka om det gick att urskilja någon effekt på upplevelsen av innehållet i en copytext i form av trovärdighet och övertygelse, beroende på om skribenten var ChatGPT eller en mänsklig copywriter, samt om det gick att urskilja någon effekt på upplevelsen av skribenten i form av kreativitet och professionalitet. För att uppnå studiens syfte formulerades följande frågeställningar: 1. Hur trovärdig upplevs en AI-genererad text jämfört med en text skriven av en mänsklig copywriter? 2. Hur övertygande upplevs en AI-genererad text jämfört med en text skriven av en mänsklig copywriter? 3. Hur trovärdig upplevs ChatGPT jämfört med en mänsklig copywriter? 4. Finns det ett samband mellan attityden till AI och upplevelsen av innehållet? Studiens teoretiska ramverk bestod av teori och tidigare forskning gällande generativ artificiell intelligens (GAI), rekommendationer för att skriva en prompt, attityder till AI, copywritingkompetensen och rekommendationer för att skriva en effektiv copytext. De teoretiska begreppen trovärdighet och övertygelse är två centrala begrepp för studien som har operationaliserats till kompetens och rekommenderat innehåll samt unique selling proposition (USP), call to action (CTA), retorik och historieberättande. Resultaten visar att de som fick läsa den AI-genererade copytexten hade en mer positiv upplevelse av innehållet och skribenten än de som fick läsa den mänskligt skrivna copytexten. Resultaten visar däremot att kreativiteten hos skribenten upplevdes lika hög av de båda experimentgrupperna. Det fanns överlag inga statistiskt signifikanta samband mellan upplevelsen av innehållet och attityden till AI. Utifrån resultatens signifikans och dess medelvärden landar studiens slutsatser i att texten skriven av ChatGPT upplevs vara mer trovärdig i form av kompetens och rekommenderat innehåll, än texten skriven av den mänskliga copywritern och att texten skriven av ChatGPT upplevs vara mer övertygande i form av USP, CTA, retorik samt historieberättande, än texten skriven av den mänskliga copywritern. Vi kan även konstatera att ChatGPT upplevs vara mer trovärdig i form av kompetens som operationaliserats till professionalitet, än texten skriven av den mänskliga copywritern, vilket däremot skulle kunna bero på slumpen då resultatet inte var statistiskt signifikant. Sammanfattningsvis kan vi utifrån studiens resultat dra slutsatsen att ChatGPT hotar copywritingkompetensen. Däremot är det inte AI-verktyget i sig som hotar yrkesrollen, då det i nuläget fortfarande kräver mänsklig involvering. Det är i stället kompetensen inom prompt engineering som hotar yrkesrollen, då denna kunskap enligt studien är avgörande för att åstadkomma en trovärdig och övertygande AI-genererad copytext.

Text-to-image generation is one of the most significant technological advancements in artificial intelligence (AI) during recent years, with the software Midjourney being amongst the most popular tools for this purpose. There is however some uncertainty about how tools like Midjourney will impact and integrate with creative professions, such as graphic design. To successfully integrate graphic designers with image-generating software, it is necessary for the tools to be designed according to the needs of their design process. This study, therefore, examines the requirements graphic designers have for Midjourney to be integrated into the idea generation phase of their design process. Requirements were gathered through interviews, which were then prioritized through surveys. A comprehensive list of requirements was formulated based on these findings. An analysis of the results was also conducted to analyze the prioritization of each individual requirement. The results highlight areas of improvement in Midjourney, with a focus on control and usability. The discussion explores factors that could have influenced the validity of the results, along with a reflection on the study's implementation. Finally, the conclusion of the study outlines the key improvements needed for Midjourney according to the results and suggests directions for future research within the research area.

This thesis addresses the intersection of reception history in biblical studies, Generative Artificial Intelligence (GAI) and phenomenology. Three images, from text prompts using different English translations of Mark 1:1–8 (KJV, NRSV and NIV) have been generated by GAI. In addition to the three translations, a more encompassing body of information, based on exegetical analysis, reception history and recent scholarly literature on John the Baptist and Mark 1, was also provided. Mark 1 is analyzed using narrative criticism with special focus on John the Baptist. Current research on the historical John is discussed, alongside interpretations of John from Late Ancient Christian Sources seen from a phenomenological perspective.  Traditionally, interpreting biblical art and text has assumed an artist portraying a narrative reading using methods such as visual exegesis. With GAI, this has changed moving the artist from the canvas to the text prompt. It puts the biblical text in a direct causal connection to the created image. Previously the artist had to decide when the image was finished but with GAI the decision is about which image to keep. The purpose of the image becomes a focal point. Images created with this modern technology can be relevant in at least two regards. First, they do represent a new type of biblical art. Second, the iterative process itself is a novel approach to studying and interacting with the Bible. Challenges exists, such as a bias towards Western/American cultural, sociological, and economical values. Data scientists and mathematicians are determining the probabilistic models without problematizing the content. Ethical questions in this field need to be addressed. GAI learning from AI-produced data – instead of human data – will likely become an issue, thus reinforcing existing biases and prejudices further.