Chatbots’ emergence into education raises critical questions about the nature of writing in the digital age. This study-in-progress explores upper secondary school students’ interactions with chatbots during writing processes in the Norwegian subject.

Writing is both a subject-specific and cross-curricular competence. In Norway, the Norwegian subject has a particular responsibility for fostering students’ writing skills (Utdanningsdirektoratet, 2020). The emergence of generative artificial intelligence (AI), particularly large language models (LLMs) and chatbots, challenges traditional conceptions of writing, prompting the need for empirical research on how students engage with these tools.

Conversational chatbots, like ChatGPT, simulate and generate human-like texts (Lo, 2023). Importantly, they are part of students’ every day and school-life (Holm et al., 2025) yet their role in writing instruction remains underexplored. Understanding students’ interactions with chatbots is crucial for informing pedagogy, subject didactics, curriculum development and policy.

Research on chatbots is emerging, primarily focusing on higher education (Sørhaug, 2024). Studies examine teachers’ and students’ perceptions of chatbots as writing assistants, highlighting benefits for writing support, alongside concerns about plagiarism and academic integrity (e.g.Karataş et al., 2024; Zhao et al., 2024). Barrett and Pack (2023) found students and teachers considered chatbots more acceptable for early stages of writing (e.g. brainstorming) than later in the writing process. Research on feedback on writing showed that students were divided in preferring chatbot-generated or human feedback (Escalante at al., 2023). These studies are insightful yet reveal a gap in the empirical research on students’ actual interactions with chatbots during writing.

A few studies have explored students’ use of chatbots, but focused on specialized chatbots or outside school contexts. Guo et al. (2024) explored undergraduate students’ using a chatbot designed for argumentative essays, finding they had to employ other mediating tools (e.g. typing tools, online resources) to facilitate their interaction. Levine et al. (2024) analyzed high school students’ use of ChatGPT in an after-school setting, finding that students primarily used it for planning and proofreading, but rarely adopted ChatGPT’s exact language. Both studies underscore the need to investigate further in broader classroom settings. Although projects are emerging in Norway, research on chatbots and writing is at an initial stage. Nationally and internationally qualitative research on school-aged students using chatbots is scarce (Sørhaug, 2024). Therefore, this study asks the following research question (RQ):  

RQ: What characterizes students’ interactions with chatbots during writing processes in the Norwegian classroom?

The study adopts sociocultural perspectives on learning, writing and technology. Drawing on Vygotsky (1978) and Wertsch (1991; 1998) writing is conceptualized as a socially mediated activity in which students use mediational means to construct meaning.

Mediational means include material tools (e.g. word processors) and symbolic signs (e.g. written language) that shape and transform human action (Wertsch, 1998). However, humans do not passively adopt, but also play an active role in using and transforming these means (Wertsch, 1998). In other words, there is an irreducible tension between humans and the mediational means they employ (Wertsch, 1998).

Chatbots may, however, blur the boundaries between humans and mediational means, as they are both material tools (algorithms, statistical models), but also signs, built on human language and generating human-like language. This dual nature raises questions about agency in writing. Students may use chatbots as co-authors or collaborators or authorities that influence their writing. Or simply as tools for quick text generation, to be copied and pasted.

By applying a sociocultural lens, the study investigates students’ interactions with chatbots as mediational means, exploring how chatbot-generated text is shaped by and may in turn shape students' writing practices. This will contribute to providing empirical insights into how emerging AI technologies influence writing development and pedagogy.

The study takes on a qualitative case study approach, understood as an in-depth description and analysis of a bounded system within its real-life context (Merriam & Tisdell, 2016). The project has a longitudinal dimension as I follow the same participants over three years. Data for this study is from year one. Participants are two Norwegian teachers and 11 of their first-year students at an upper secondary school. In Norway, students attend upper secondary for three years and participants have agreed to take part during this period. Participation is voluntary, based on informed consent and the project adheres to national guidelines for research ethics (NESH, 2021). Primary data consists of video recordings of the participants when engaging in a writing task in class. The main objective is to explore students’ interactions with chatbots under as “naturally occurring” circumstances as possible (Levine et al., 2024). Two types of video data was collected during a two-hour class session in November 2024: 1) screencasts of each student’s screen and 2) regular video recordings of the students. In total, approximately 44 hours, i.e 4 hours of recodrings per student (2 hours screencast, 2 hours video) Previous research deems video-data as providing opportunities to systematically look for patterns and thoroughly examine multifaceted, complex social practices that would be impossible to observe in situ (Blikstad-Balas, 2017). Thus, the collected data for this study offers comprehensive and rich opportunities for in-depth analysis.. Secondary types of data have been collected, inspired by ethnographic approaches, like interviews, field notes and relevant documents. These will not be subject to independent analysis but provide important contextual understandings. Video data is analyzed through interaction analysis (IA) (Jordan & Henderson, 1995). IA is a valuable and relevant tool for analyzing complex real-world settings and a suitable to empirically investigate the interactions of human beings with each other and with objects in their environment (Jordan & Henderson, 1995). Through IA I focus on capturing nuances of real-time communication and interactions between students and chatbots in particular, but I am also able to see other mediational means they employ. As this is work in progress, I have started the analytical work by creating content lists and logs of the videos. This is in line with IA as the first step of the analysis to make up the basis for choosing and going in-depth of particularly interesting segments or themes across the data (Jordan & Henderson, 1995).

The expected outcomes of this study are to provide empirical insights into how students use, adapt to, and are influenced by chatbots in the writing process. The findings will contribute to discussions on the evolving role of writing in the age of chatbots, and inform educators, policymakers, and researchers. Although this is a study in progress and data analysis has not yet been completed, preliminary findings indicate significant variation in students’ chatbot practices. Students employ different strategies, make distinct choices and display differing levels of competencies in interacting with the chatbot. This is for example visible in how they formulate prompts or choose which parts of the generated text to use or not. These variations might be linked to students’ overall writing skills and strategies, suggesting that using chatbots might be less beneficial for struggling writers, while proficient writers are able to use chatbots more effectively. Despite differences, some common patterns also seem to emerge from the data. 1) Fairly limited use of chatbots during the writing process: Many students seem to draw more on traditional and familiar resources, such as school websites, word processing software, paper documents, or asking the teacher for help while writing. 2) Minimal conversational engagement with the chatbots: While the students use prompts to generate text, they rarely follow up with additional questions or ask for clarifications or further explanations. Instead, chatbots function more like search engines than interactive writing partners. 3) Concerns about cheating and academic integrity: Students express uncertainty about whether using chatbots is cheating or not. This concern surfaces in discussions with teachers and peers, as well as through behavior, like using AI detection tools to assess the originality of their own writing. More detailed findings, discussions and implications will be in place and presented at the conference.

Barrett, A., & Pack, A. (2023). Not quite eye to AI: student and teacher perspectives on the use of generative artifcial intelligence in the writing process. International Journal of Educational Technology in Higher Education(20), 61. Blikstad-Balas, M. (2017). Key challenges of using video when investigating social practices in education: Contextualization, magnification, and representation. International journal of research & method in education, 40(5), 511-523. Guo, K., Li, Y., Li, Y., & Chu, S. K. W. (2024). Understanding EFL students’ chatbot-assisted argumentative writing: An activity theory perspective. Education and Information Technologies, 29(1), 1-20. https://doi.org/10.1007/s10639-023-12230-5 Holm, N.-K. T., Neuhaus, S. V., & Bitsch, L. (2025). De unges stemme i samtalen om generativ AI. https://videnogdemokrati.dk/app/uploads/2025/01/DEMX.AI-rapport.final-01a_rapport.pdf?ref=viden.ai Jordan, B., & Henderson, A. (1995). Interaction analysis: Foundations and practice. The journal of the learning sciences, 4(1), 39-103. Karataş, F., Abedi, F. Y., Ozek Gunyel, F., Karadeniz, D., & Kuzgun, Y. (2024). Incorporating AI in foreign language education: An investigation into ChatGPT’s effect on foreign language learners. Education and Information Technologies, 1-24. Levine, S., Beck, S. W., Mah, C., Phalen, L., & PIttman, J. (2024). How do students use ChatGPT as a writing support? Journal of Adolescent & Adult Literacy. Lo, C. K. (2023). What is the impact of ChatGPT on education? A rapid review of the literature. Education Sciences, 13(4), 410. Merriam, S. B., & Tisdell, E. J. (2016). Qualitative research: A guide to design and implementation (4th ed.). John Wiley & Sons. NESH, D. n. f. k. f. s. o. h. (2021). Forskningsetiske retningslinjer for samfunnsvitenskap og humaniora. De nasjonale forskningsetiske komitéene Retrieved from https://www.forskningsetikk.no/retningslinjer/hum-sam/forskningsetiske-retningslinjer-for-samfunnsvitenskap-og-humaniora/ Sørhaug, J. O. (2024). "Takk for en flott diskusjon!" Med chatboten Sokrates som samtalepartner i tre elevgruppers fagsamtalar. In K. Kverndokken & J. O. Bakke (Eds.), 101 måter å fremme muntlige ferdigheter på - en teoretisk og praktisk muntlighetsdidaktikk (pp. 69-96). Fagbokforlaget. Utdanningsdirektoratet. (2020). Læreplan i norsk (NOR01-06). Læreplanverket for kunnskapsløftet Retrieved from https://www.udir.no/lk20/nor01-06 Vygotsky, L. S. (1978). Mind in society: Development of higher psychological processes. Harvard university press. Wertsch, J. V. (1991). Voices of the mind : a sociocultural approach to mediated action. Harvard University Press. Wertsch, J. V. (1998). Mind as action. Oxford university press. Zhao, X., Cox, A., & Cai, L. (2024). ChatGPT and the digitisation of writing. Humanities and Social Sciences Communications, 11(1), 1-9.