What is your professional background and primary area of research?
"I am a researcher at the Danish School of Education (DPU), Aarhus University. My research focuses on cultural learning processes: how people learn in and through culture, and how materials and technologies (e.g., AI) are incorporated into learning and practice. I work in the field of educational anthropology and have developed an “anthropology of learning” about how we as participants are shaped by - and shape - cultural ecologies through frictions in everyday practice. This field ties in with my previous work on interpretive frameworks, cultural resources, and human-technology relations."

What is your connection to SHAPE, and what is your research project at SHAPE about?
"The project “Socratic ignorance in working with generative AI chatbots” was funded by Aarhus University's SHAPE initiative. In the project, we investigated how students actually use ChatGPT (3.5) in teaching-related situations and how their prior learning and academic interpretive frameworks influence their prompting and learning outcomes. We developed the concept of Relational Socratic Ignorance (RSI): a pedagogical awareness that ignorance is always relational to an academic framework - and that both questions and answers to AI must therefore be framed academically. Methodologically, we conducted a teaching experiment (including the task “What is education?”) and showed that some students challenged ChatGPT's instrumental definitions by drawing on broader academic resources, while others lacked such resources to activate."

What impact do you expect your project to have on society or your field of research?
"Three main contributions stand out: (1) A more nuanced view of human-machine relations in education; (2) a theoretical and practical emphasis on the importance of prior learning and cultural resources when students select and qualify AI responses; (3) a concrete pedagogical concept (RSU) that can be used to frame prompting and strengthen critical thinking, so that AI becomes an academic tool rather than an answer machine. Together, these can inform didactics, guidance, and testing methods in higher education, thereby influencing both practice and research in learning and technology."

What future projects do you have in the pipeline?
"The article marks an early, qualitative study, which we will follow up with larger and more varied practical studies. Among other things, we are working on (a) testing RSU across subjects and educational programs, (b) developing teaching designs and teaching competencies that operationalize RSU in specific courses, and (c) investigating how RSU can mitigate differences in students' prerequisites when they integrate AI into their studies. The project line builds on our pedagogical-anthropological theory of cultural resources and learning potentials and explicitly invites more empirical, practice-oriented research in the field.

Our SHAPE research shows that generative AI is not culturally neutral: models such as ChatGPT reflect the linguistic and educational interpretative frameworks on which they are trained - for example, the Danish concept of dannelse (cultivation) disappears unless explicitly prompted, because the model is primarily rooted in American traditions. At the same time, there is a risk of inequality between students when the technology overlooks the importance of students' preceding learning: concepts and materials already acquired are crucial to what one can see, ask about, and learn in new situations—these differences in preceding learning create differences in learning opportunities, which RSU seeks to address didactically.

The protocol behind Pedagogical Insight documents that the field is dominated by surveys, technology-driven/"teacher-bot" studies, and non-empirical recommendations, while ethnographic in situ studies of students' actual use are significantly underrepresented—despite a screening of nearly 14,000 articles from 2018 to spring 2024. This is precisely why our SHAPE work is unique: we investigate how students actually work with GAI in practice and show that the technology is culturally shaped and that prior learning structures what they can get out of it – exactly the type of practice-oriented insight that the protocol calls for in SSH education programmes."