The IT University of Copenhagen (ITU) announced DKK 7.1 million in support from Independent Research Fund Denmark on 9 November 2026 for a five-year educational AI project. Led by Jichen Zhu, the research will investigate ambiguity in learning supported by large language models. Its planned chatbot, Diffrakt, is a research output to be developed.
What the Diffrakt project will investigate
ITU’s announcement names the project Friction by Ambiguity: Improving Engagement and Critical Thinking in LLM-Supported Learning. Diffrakt will use guided questions and deliberately open responses in AI ethics teaching. The team will also research how to calibrate ambiguity to model uncertainty.
Christian Hardmeier is co-principal investigator, with Pawel Grabarczyk contributing ethics expertise. Initial testing is planned with ITU students, alongside work with educators. These are research plans; the announcement does not establish that Diffrakt improves learning.
Research and funder context
Zhu’s institutional profile identifies Zhu as a professor in Human-Computer Interaction and Design and head of the Human-AI Interaction Lab. Zhu’s research includes learning technologies. That places the project within human-computer interaction research as well as AI development.
The lab’s description connects interaction design and critical technical practice with research on how people interpret and live with AI. Learning is among its research domains.
Independent Research Fund Denmark describes its remit as supporting researcher-initiated, high-risk research within and across disciplines, alongside thematic funding.
What universities and research offices can take from it
For institutions considering educational AI, the useful distinction is between a funded research question, a planned tool and evidence of educational benefit. November 9 marks ITU’s announcement; it should not be treated as the funder’s grant decision date.
An institutional assessment can use that distinction to frame its research evaluation: track the proposed technology separately from evidence about students’ reasoning and engagement. Funding and a chatbot development plan alone cannot establish that the educational approach works.
Sources
- ITU: DKK 7.1 million for ITU research into a new generation of educational AI. Checked 2026-11-14.
- IT University of Copenhagen: Jichen Zhu profile. Checked 2026-11-14.
- IT University of Copenhagen: Human-AI Interaction Lab. Checked 2026-11-14.
- Independent Research Fund Denmark: About the fund. Checked 2026-11-14.







