Mari Serine Kannelønning

• Muir, Tom & Kannelønning, Mari Serine (2026). AI as an Academic Writing Companion: Prompting, Rhetoric and Research Writing. Journal of Academic Writing. 16(1), s. 1–14. doi: 10.18552/joaw.v16i1.1481. Fulltekst i vitenarkiv Vis sammendrag

  This Teaching Practice paper documents and reflects on a workshop aiming to infuse rhetorical knowledge into Gen AI prompting. The workshop – “AI as an Academic Writing Companion” – is aimed at PhD candidates and academic faculty. It runs at a large Scandinavian university where English is a second language for most researchers, but is very much the expected language of academic publication. It is easy to be dazzled by the superficial, sentence-level fluency of Generative AIs, especially if one is a novice researcher writing in a second language. This superficial fluency may obscure the AI’s lack of rhetorical principles, however: the AI may lack them, and the researcher may not know the AI lacks them. To combine academic literacy with a critical sense of the capabilities of AI chatbots, therefore, the workshop draws on Moxley’s idea of kairos-based prompting. The aim here is to develop and apply the rhetorical knowledge necessary to guide a bot’s responses; in this way, the workshop combines the needs of researchers for both traditional academic literacies and generative AI literacy. The article describes the prompts and the pedagogy behind them, and our reflections on the development of the workshop.

• Kannelønning, Mari Serine & Grisot, Miria (2026). Collective efforts for the introduction of AI in healthcare: An empirical study of two Norwegian initiatives. Tidsskrift for forskning i sygdom og samfund. ISSN 1604-3405. 25(44), s. 100–123. doi: 10.7146/tfss.v25i44.156499. Fulltekst i vitenarkiv Vis sammendrag

  The collective dimension of AI is crucial to ensure sustainable and ethical outcomes of AI implementation in society; accordingly, the introduction of AI in healthcare is a collective concern. In this paper, we examine this collective dimension of AI introduction by drawing on an empirical longitudinal case study of two initiatives aimed at introducing AI into the Norwegian public healthcare services. The first initiative is an inquiry process led by the Norwegian Directorate of Health, as part of a project initiated by a government policy plan outlining a vision for AI in healthcare. The second initiative is a nationwide, informally established network of professionals engaged with AI in healthcare. Drawing on insights on hybrid knowledge production in collective efforts, as well as practice-oriented and relational understandings of technology, our analysis of the two initiatives focuses on how the collective effort was carried out, the role of the orchestrators and the implications of such efforts on their outcomes. Our findings show that the two initiatives, while seeking to produce hybrid knowledge on how to best introduce AI in healthcare, faced different challenges, both in generating such knowledge, in achieving consensus-based results and in enacting visions of AI in healthcare oriented to clinical practice. We suggest that to contribute to a consensus-based and inclusive introduction of AI in healthcare, collective efforts should pay attention to the inclusion of heterogeneous stakeholders and their interaction modes, the politics of agenda settings and modes of orchestration.

• Soetens, Esli; Kannelønning, Mari Serine & Grisot, Miria (2025). "Frictions In AI Implementation In Specialised Norwegian Healthcare: An interview-based study". I Li, Magnus; Hylving, Lena & Mønsted, Troels Sune (Red.), The proceedings of the 16th Scandinavian Conference on Information Systems (SCIS) held in Oslo, Norway, 2025. Association for Information Systems (AIS). ISSN 9781958200148. doi: https:/aisel.aisnet.org/scis2025/3. Fulltekst i vitenarkiv Vis sammendrag

  AI technologies are envisioned as a solution to current societal challenges, including in the healthcare sector. However, their implementation in real-world hospital settings remains limited. In this paper we examine the slow implementation of AI in healthcare as a problem of installed base constraints. AI technologies are fundamentally different from traditional medical information technology. However, existing implementation practices are still shaped by regulations, development and implementation approaches established for conventional medical information technologies, constituting the installed base. We examine these constraints as frictions between novel AI technologies and the installed base through an interview-based study with 14 interviews with 11 experts in the Norwegian healthcare sector. We address the following research question: How do frictions between novel AI technologies and the existing sociotechnical installed base shape the process of AI implementation in public hospitals? Our findings reveal seven key frictions relating to data complexities, hospital information infrastructures, regulatory landscape, resource requirements, accountability, implementation strategies, and incentive structures. We contribute to understanding the shift needed to support the implementation and use of AI technologies in healthcare.

• Vikan, Magnhild; Aryan, Ramtin; Kannelønning, Mari Serine; Riegler, Michael & Danielsen, Stein Ove (2025). Reflecting on LLM Support in Reflexive Thematic Analysis: An Exploratory Study. Qualitative Health Research. ISSN 1049-7323. doi: 10.1177/10497323251365211. Fulltekst i vitenarkiv Vis sammendrag

  The launch of ChatGPT in November 2022 accelerated discussions and research into whether base large language models (LLMs) could increase the efficiency of qualitative analysis phases or even replace qualitative researchers. Reflexive thematic analysis (RTA) is a commonly used method for qualitative text analysis that emphasizes the researcher’s subjectivity and reflexivity to enable a situated, in-depth understanding of knowledge generation. Researchers appear optimistic about the potential of LLMs in qualitative research; however, questions remain about whether base models can meaningfully contribute to the interpretation and abstraction of a dataset. The primary objective of this study was to explore how LLMs may support an RTA of an interview text from health science research. Secondary objectives included identifying recommended prompt strategies for similar studies, highlighting potential weaknesses or challenges, and fostering engagement among qualitative researchers regarding these threats and possibilities. We provided the interview file to an offline LLM and conducted a series of tests aligned with the phases of RTA. Insights from each test guided refinements to the next and contributed to the development of a recommended prompt strategy. At this stage, base LLMs provide limited support and do not increase the efficiency of RTA. At best, LLMs may identify gaps in the researchers’ perspectives. Realizing the potential of LLMs to inspire broader discussion and deeper reflections requires a well-defined strategy and the avoidance of misleading prompts, self-referential responses, misguiding translations, and errors. Conclusively, high-quality RTA requires a human, comprehensive familiarization phase, and methodological competence to preserve epistemological integrity.

• Severinsen, Gro-Hilde; Pedersen, Rune; Ellingsen, Gunnar; Graff, Bjørn Anton; Tveiten, Line & Kannelønning, Mari Serine [Vis alle 7 forfattere av denne artikkelen] (2025). Implementing and Scaling Commercial AI in Public Healthcare Settings: Experiences from Norway. Studies in Health Technology and Informatics. ISSN 0926-9630. 326, s. 7–11. doi: 10.3233/SHTI250226. Fulltekst i vitenarkiv
• Silsand, Line; Kannelønning, Mari Serine; Severinsen, Gro-Hilde & Ellingsen, Gunnar Adelsten (2024). Enabling AI in Radiology: Evaluation of an AI Deployment Process. Studies in Health Technology and Informatics. ISSN 0926-9630. 316, s. 580–584. doi: 10.3233/SHTI240480. Fulltekst i vitenarkiv Vis sammendrag

  Artificial intelligence (AI) is expected to transform healthcare systems and make them more sustainable. Despite the increased availability of AI tools for disease detection, evidence of their impact on healthcare organisations and patient care remains limited. Drawing on previous research underscoring the need for comprehensive evaluations of real-world AI deployments, this paper explores the challenges and opportunities encountered while procuring and implementing AI solutions for radiology. The paper aims to contribute to a better understanding of the complexities surrounding AI deployments in real-world clinical settings through a process evaluation study.

• Kannelønning, Mari Serine; Grisot, Miria & Williams, Robin (2024). Towards Experimental Implementations: Moving Emerging Artificial Intelligence Technologies into Real-World Clinical Settings. I Miele, Francesco & Giardullo, Paolo (Red.), Reframing Algorithms. STS perspectives to Healthcare Automation. Palgrave Macmillan. ISSN 9783031520495. s. 147–172. doi: 10.1007/978-3-031-52049-5_7. Fulltekst i vitenarkiv
• Williams, Robin; Anderson, Stuart; Cresswell, Kathrin; Kannelønning, Mari SerinE; Mozaffar, Hajar & yang, Xiao (2024). Domesticating AI in medical diagnosis. Technology in society. ISSN 0160-791X. 76. doi: 10.1016/j.techsoc.2024.102469. Fulltekst i vitenarkiv Vis sammendrag

  We consider the anticipated adoption of Artificial Intelligence (AI) in medical diagnosis. We examine how seemingly compelling claims are tested as AI tools move into real-world settings and discuss how analysts can develop effective understandings in novel and rapidly changing settings. Four case studies highlight the challenges of utilising diagnostic AI tools at differing stages in their innovation journey. Two ‘upstream’ cases seeking to demonstrate the practical applicability of AI and two ‘downstream’ cases focusing on the roll out and scaling of more established applications. We observed an unfolding uncoordinated process of social learning capturing two key moments: i) experiments to create and establish the clinical potential of AI tools; and, ii) attempts to verify their dependability in clinical settings while extending their scale and scope. Health professionals critically appraise tool performance, relying on them selectively where their results can be demonstrably trusted, in a de facto model of responsible use. We note a shift from procuring stand-alone solutions to deploying suites of AI tools through platforms to facilitate adoption and reduce the costs of procurement, implementation and evaluation which impede the viability of stand-alone solutions. New conceptual frameworks and methodological strategies are needed to address the rapid evolution of AI tools as they move from research settings and are deployed in real-world care across multiple settings. We observe how, in this process of deployment, AI tools become ‘domesticated’. We propose longitudinal and multisite ‘biographical’ investigations of medical AI rather than snapshot studies of emerging technologies that fail to capture change and variation in performance across contexts.

• Kannelønning, Mari Serine (2023). Contesting futures of Artificial Intelligence (AI) in healthcare: formal expectations meet informal anticipations. Technology Analysis & Strategic Management. ISSN 0953-7325. doi: 10.1080/09537325.2023.2226243. Fulltekst i vitenarkiv Vis sammendrag

  Artificial Intelligence (AI) technologies are expected to play an essential role in future healthcare systems for saving resources, improving treatment quality and enhancing patient safety. Governments worldwide are preparing for such a future by developing strategies and policies. Drawing on concepts and perspectives from Science and Technology Studies (STS), this article explores how government policies set the agenda for introducing AI in healthcare, using the Norwegian National Health and Hospital Plan 2020–2023 (NHSP) as an example. The article further investigates how the formal expectations included in the NHSP are met with informal anticipations expressed by actors working closer to clinical practices involved in an inquiry process initiated by the policy. Taking a qualitative approach, the article explores what characterises formal expectations of AI in healthcare and how these expectations are contested. The study finds that there are tensions between the different assumptions and that crucial issues concerning the future usage of AI are not yet on the Norwegian Government’s agenda. Pairing the forces of the formal expectations with the ambiguous concept of AI, the current chasm between AI development and deployment, and the uncertainties constituted by the contesting elements, the article concludes by introducing a paradox of inevitability.

• Silsand, Line; Severinsen, Gro-Hilde & Kannelønning, Mari Serine (2023). Preparing for Implementing Commercial Algorithms in Radiology: A Formative Evaluation Study. Reports of the European Society for Socially Embedded Technologies. doi: 10.48340/ihc2023_p010. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2023). Navigating uncertainties of introducing artificial intelligence (AI) in healthcare: The role of a Norwegian network of professionals. Technology in society. ISSN 0160-791X. 76. doi: 10.1016/j.techsoc.2023.102432. Fulltekst i vitenarkiv Vis sammendrag

  Artificial Intelligence (AI) technologies are expected to solve pressing challenges in healthcare services world- wide. However, the current state of introducing AI is characterised by several issues complicating and delaying their deployments. These issues concern topics such as ethics, regulations, data access, human trust, and limited evidence of AI technologies in real-world clinical settings. They further encompass uncertainties, for instance, whether AI technologies will ensure equal and safe patient treatment or whether the AI results will be accurate and transparent enough to establish user trust. Collective efforts by actors from different backgrounds and af- filiations are required to navigate this complex landscape. This article explores the role of such collective efforts by investigating how an informally established network of professionals works to enable AI in the Norwegian public healthcare services. The study takes a qualitative longitudinal case study approach and is based on data from non-participant observations of digital meetings and interviews. The data are analysed by drawing on perspectives and concepts from Science and Technology Studies (STS) dealing with innovation and socio- technical change, where collective efforts are conceptualised as actor mobilisation. The study finds that in the case of the ambiguous sociotechnical phenomenon of AI, some of the uncertainties related to the introduction of AI in healthcare may be reduced as more and more deployments occur, while others will prevail or emerge. Mobilising spokespersons representing actors not yet a part of the discussions, such as AI users or researchers studying AI technologies in use, can enable a ‘stronger’ hybrid knowledge production. This hybrid knowledge is essential to identify, mitigate and monitor existing and emerging uncertainties, thereby ensuring sustainable AI deployments.

• Kannelønning, Mari Serine & Evjen, Sunniva (2018). Tillit og tverrfaglighet - om bibliotekaren som forskerstøtte. Nordisk Tidsskrift for informationsvitenskab- og kulturformidling. ISSN 2245-2931. 7(3), s. 74–88. doi: 10.7146/ntik.v7i3.111488. Fulltekst i vitenarkiv Vis sammendrag

  Technological advancements and reforms in the university sector have changed the conditions under which researchers and academic staff work. In turn, university libraries need to adapt to the changing needs of this particular user group. This article highlights which competencies and ro/es the university librarian needs in order to address these changing needs based on an understanding of the researcher's work context. How can the academic librarian contribute to the research process? The study on which this article is based has applied lnstitutional Ethnography {IE}, as one methodological approach. IE puts focus on the expert knower, in this case the researcher, who can transmit work knowledge specific to the context in which he or she operates. Applying this approach is one way of dealing with cha/lenges concerning tacit know/edge. In addition, IE provides an opening to revealing new knowledge about a we/1-studied issue. The data material consists of interviews with researchers (individual and group), and a workshop with academic library staff. Findings show that the researchers' demands of publications governs their prioritization, but also their relationships to technology and people. The informants' state that they need to be ab/e to trust that librarians contribute added va/ue and substance to the research process - only then can they be part of the research network. The informants wou/d also like librarians to contribute with a transdisciplinary mindset in transdisciplinary research projects, that they coordinate administrative support, and take charge of research dissemination intended to "popularize" research.

Se alle arbeider i NVA

• Kannelønning, Mari Serine (2026). KI i spesialisthelsetjenesten - fra strategi til praksis. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine; Kempton, Alexander Moltubakk; Mønsted, Troels Sune & Grisot, Miria (2026). Information Infrastructure governance under urgency. Dealing with uncontrolled genAI adoption in Norwegian hospitals. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Lingjærde, Ole Christian (2026). KI i helsetjenesten: på tide å skifte fokus? Fulltekst i vitenarkiv Vis sammendrag

  Veien mot gevinstrealisering med KI vil starte med at beslutningstakere på ulike nivåer skifter fokus og i høyere grad ser på KI-innføring som noe mer enn et spørsmål om teknologiens pris, egenskaper og effektiviseringspotensial.

• Kannelønning, Mari Serine (2026). Expectations, promises and (hybrid) knowledge production: The shaping of a future with AI in healthcare. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2026). KI i spesialisthelsetjenesten: hva vil det si og hvor er vi på vei? Fulltekst i vitenarkiv
• Muir, Tom & Kannelønning, Mari Serine (2025). AI as an Academic Writing Companion. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2025). KI i helsetjenesten: gevinster, spenninger og ledelsesinnsikter. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2025). En fremtid med KI i spesialisthelsetjenesten tar form. Forventninger, teknologiske løfter og hybrid kunnskapsproduksjon. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2025). KI i spesialisthelsetjenesten: fra visjoner til resultat av innføring. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2025). Kunstig intelligens i forskning. Fulltekst i vitenarkiv Vis sammendrag

  Stort og smått til refleksjon og diskusjon om KI-bruk i forskning.

• Kannelønning, Mari Serine & i (NRK), Sumaya Mohsini (2025). Generativ KI: Over halvparten av nordmenn bruker det i 2025. [Internet]. Nrk.no. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2025). Kunstig intelligens - hva er det og hvordan brukes det? Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2025). KI som forskningsassistent – om (ansvarlig) KI-bruk forskning. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Ørneseidet, Thea (2025). Hvilke forventninger har vi til KI i norske sykehus? [Radio]. Viten og snakkis - en forskningspodkast. doi: https:/open.spotify.com/episode/6JOkApud3x6O2MoTeDbSeT?si=dcdb4ba609a840dc. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Tallerås, Kim (2025). Navigating Grey Zones: Ethical Boundaries of AI Use in Academia. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Grisot, Miria (2025). Collective preparation work for enabling algorithmic organising in healthcare. Fulltekst i vitenarkiv
• Severinsen, Gro-Hilde; Pedersen, Rune; Ellingsen, Gunnar; Kannelønning, Mari Serine & Silsand, Line (2025). Comparing a Top-Down and a Bottom-Up Approach for Implementing AI in Radiology Practice. Studies in Health Technology and Informatics. ISSN 0926-9630. 327, s. 424–425. doi: 10.3233/SHTI250368. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine; Gasparini, Andrea Alessandro & Ringnes, Hege Kristin (2024). Vi må snakke mer om KI-bruk i forskning. Fulltekst i vitenarkiv
• Soetens, Esli; Kannelønning, Mari Serine & Grisot, Miria (2024). Exploring approaches to medical AI deployment in Norwegian hospitals: In-house development and commercial procurement. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2024). KI i helsetjenestene. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Kvalnes, Pål Arne (2024). KI på norske sykehus: – Det kommer til å bli annerledes enn vi tror. [Internet]. oslomet.no. Fulltekst i vitenarkiv
• Ringnes, Hege Kristin; Kannelønning, Mari Serine & Gasparini, Andrea Alessandro (2024). AI and emotions: researchers' emotional reactions when interacting with AI tools. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2024). KI-bruk og vitenskapelig publisering. For forfattere, fagfeller og redaktører. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Ringnes, Hege Kristin (2024). Kunstig intelligens (KI) som forskningsassistent: muligheter og utfordringer. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2024). Kunstig intelligens: Potensiale og begrensninger for intern og brukerorientert biblioteksvirksomhet. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2024). KI i helse: visjoner, dagens status og fremtidsutsikter. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Hanssen, Anders Braarud (2024). Skal vi vinne i KI-kappløpet, må vi gå motstrøms. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2024). KI: – Det er ikke teknologien som er vanskelig – det er å endre arbeidsvaner. [Journal]. Dagens medisin. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2024). Forskernatt - forskernes eget talkshow. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Hanssen, Anders Braarud (2024). Skal vi vinne i KI-kappløpet må vi gå motstrøms. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine; Soetens, Esli & Grisot, Miria (2024). Exploring multimodal AI – scaling AI infrastructures in healthcare. Fulltekst i vitenarkiv
• Seibt, David; Anderson, Stuart; Elizondo, Andrey; Kannelønning, Mari Serine; Williams, Robin & Yang, Xiao (2024). Competing Models of Platformization for Delivering Health AI. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine; Silsand, Line & Severinsen, Gro-Hilde (2024). Bærekraftig KI-implementering på norske sykehus. Fulltekst i vitenarkiv
• Silsand, Line; Severinsen, Gro-Hilde & Kannelønning, Mari Serine (2023). Preparing for Implementing Commercial Algorithms in Radiology: A Formative Evaluation Study. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2023). Mirakelkur eller ei – kunstig intelligens må tas i bruk for å finne svaret. Fulltekst i vitenarkiv Vis sammendrag

  https://www.dn.no/innlegg/kunstig-intelligens/helse/sykehus/mirakelkur-eller-ei-kunstig-intelligens-ma-tas-i-bruk-for-a-finne-svaret/2-1-1509233

• Kannelønning, Mari Serine (2023). Vi kan ikke bare være føre var med den kunstige intelligensen, heller ikke i helsevesenet. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2023). KI-milliarden må brukes til forskning på forståelse og bruk. Fulltekst i vitenarkiv Vis sammendrag

  https://khrono.no/ki-milliarden-ma-brukes-til-forskning-pa-forstaelse-og-bruk/809891

• Kannelønning, Mari Serine (2022). Transforming expectations: Moving medical AIs into the world of radiology practices. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2022). Domesticating diagnostic AI (panel). Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2022). Ny forskerskole for kunstig intelligens kan gi oss det samfunnet trenger. Khrono.no. ISSN 1894-8995. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2022). Troubled terrain: Introducing AI in healthcare. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2022). Veien mot innføring av kunstig intelligens i sykehusene. Funn og refleksjoner fra studier av tre norske KI-initiativ. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2022). Kunstig intelligens og likestilling. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2022). The road towards introducing AI in Norwegian public hospitals. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2021). Panel debate: AI research and education. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2021). Kunstig intelligens i spesialisthelsetjenesten. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2019). Tid og tillit – om bibliotekutvikling og forskningsstøtte. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Ødemark, Ingjerd Legreid (2017). Hva kan et moderne UH-bibliotek være? Fulltekst i vitenarkiv
• Kannelønning, Mari Serine & Moxnes, Gry Bettina (2017). Erfaringer fra Erasmus+ utveksling. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2016). Fagbibliotekene i 2025 - Relevans, legitimitet og oppdrag. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2016). Fagbibliotekene i 2025 - relevans, legitimitet og oppdrag. Bibliotekaren (Oslo). ISSN 0804-4147. s. 12–13. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2015). Rethinking the presentation of library services. Fulltekst i vitenarkiv
• Kannelønning, Mari Serine (2023). ‘How will it hit us?’ Introducing artificial intelligence (AI) in the Norwegian healthcare services: Three modes of actor mobilisation. OsloMet - storbyuniversitetet. doi: https:/hdl.handle.net/11250/3112563. Fulltekst i vitenarkiv Vis sammendrag

  This thesis explores the introduction of artificial intelligence (AI) in the Norwegian healthcare services. It does so by focusing on the role of expectations and technological promises related to a future with AI in healthcare and how such visions are collectively pursued in the present. AI is widely expected to play an essential role in providing financially viable public healthcare services in the future. The expectations typically include promises of technologies that can automate repetitive tasks and support healthcare professionals in making decisions, thereby improving their work efficiency and enhancing treatment quality and patient safety. However, the current state of the introduction of AI in healthcare services worldwide is characterised by a so-called AI chasm between the expectations and the promise of AI technologies and their actual deployment in real-world clinical settings. To navigate this complex landscape, governments, health authorities, AI environments and other stakeholders perceive the need for a collective effort among those with relevant knowledge and expertise. This thesis uses the AI chasm as an entry point to explore three Norwegian initiatives that mobilise heterogeneous sets of actors in attempts to enable AI in the Norwegian public healthcare services: a national inquiry process led by the Norwegian Directorate of Health, an AI procurement process at a hospital trust and a nationwide network of professionals. These initiatives are studied through a qualitative case study approach that includes data from digital meeting observations, interviews and document analysis. Each study represents one of the three articles in this article-based thesis and explores different aspects of expectations, promises and collective work with the future of AI in healthcare as the focal point. Altogether, the case studies enabled an investigation of the overall research question of this thesis: What are the significance and implications of ‘mobilisation’ in the early phases of introducing ambiguous, complex and advanced technologies like AI in healthcare? The thesis analyses the findings from the case studies by drawing on concepts and perspectives from Science and Technology Studies (STS) dealing with innovation processes and the shaping of technological developments. With a primary focus on the aspect of actor mobilisation and Michel Callon’s (1986b) article ‘Some Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St Brieuc Bay’, the thesis develops a theoretical apparatus with three modes of actor mobilisation at its core: Mode 1: Steered outcome; Mode 2: Negotiated outcome; and Mode 3: Fragmented and distributed outcome. As a whole, this thesis aims to contribute to a better understanding of the role of futureoriented representations (e.g., expectations and technological promises) and the mobilisation of heterogeneous actors in the processes of introducing advanced and complex classes of technology, such as AI, into society. Denne avhandlingen tar for seg innføringen av Kunstig Intelligens (KI) i den norske helsetjenesten ved å studere ulike aktørers forventninger og løfter om fremtiden med KI i helsetjenesten, og hvordan man kollektivt jobber mot disse fremtidsvisjonene i nåtid. Verden over er det bred enighet om at KI vil spille en viktig rolle for å opprettholde økonomisk bærekraftige helsetjenester i fremtiden. Det forventes for eksempel at KIteknologier skal kunne automatisere repetitive oppgaver, støtte helsepersonell i beslutningsprosesser, forbedre arbeidseffektivitet og øke behandlingskvalitet og pasientsikkerhet. Pågående arbeid for å introdusere KI i helsetjenesten er imidlertid preget av en såkalt «KI-kløft» [AI chasm] mellom forventningene og løftene om KI-teknologier og deres faktiske implementering i virkelige kliniske praksiser. Både regjeringer, helsemyndigheter, KI-miljøer og andre interessenter mener at en felles innsats fra aktører med ulik kunnskap og ekspertise er nødvendig for å kunne navigere i dette komplekse landskapet. Avhandlingen bruker KI-kløften som inngang til å utforske tre norske initiativer som i sine forsøk på å innføre KI i norsk helsetjeneste har mobilisert heterogene aktørgrupper: en nasjonal utredningsprosess ledet av Helsedirektoratet, en KI-anskaffelsesprosess ved et større helseforetak og et landsdekkende nettverk av fagpersoner. Disse initiativene studeres gjennom en kvalitativ casestudie-tilnærming, og studien inkluderer data fra digitale møteobservasjoner, intervjuer og dokumentanalyse. Avhandlingens tre artikler tar for seg hver sin case, og utforsker ulike aspekter ved forventninger, løfter og kollektivt arbeid som har en fremtidig helsetjeneste med KI som fokusområde. Samlet utgjør disse tre studiene grunnlaget for å svare på avhandlingens overordnede forskningsspørsmål: Hva er betydningen og implikasjonene av 'mobilisering' i tidlige faser av innføringen av flertydige, komplekse og avanserte teknologier som KI i helsetjenesten? Avhandlingen analyserer funnene fra casestudiene ved å ta i bruk begreper og perspektiver fra ‘Science and Technology Studies’ (STS) som fokuserer på innovasjonsprosesser og teknologisk utvikling. Med særlig vekt på aktørmobilisering og Michel Callons artikkel: ‘Some Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St Brieuc Bay’ (1986b), utvikles det en teoretisk ramme som består av tre former for aktørmobilisering: ‘Modus 1: Styrt utfall’, ‘Modus 2: Forhandlet utfall’ og ‘Modus 3: Fragmentert og distribuert utfall’. Avhandlingens mål er å bidra til en økt forståelse for rollen til fremtidsrettede representasjoner (forventninger og løfter etc.) og mobiliseringen av heterogene aktører i innføringsprosesser av avanserte og komplekse teknologiklasser som KI i samfunnet.

Se alle arbeider i NVA