top of page

Uni Spotlight: Can AI Make Students Think Harder? Denmark Explores a New Approach to Educational AI.

Writer: A4-AI  /  (Academics for AI)
A4-AI / (Academics for AI)
Nov 11
3 min read

Updated: 5 days ago

In this article, we take a look at a 9th November 2026 press release from the IT University of Copenhagen


Eye-level view of a classroom with students engaged in learning


About the story:


A recent press release from the IT University of Copenhagen, written by Theis Duelund Jensen, announces that the university has secured DKK 7.1 million (approximately €0.95 million) in funding from Independent Research Fund Denmark for a five-year research project. The project will investigate how AI chatbots could encourage students to think more critically, rather than simply provide answers.

 

The project, titled Friction by Ambiguity: Improving Engagement and Critical Thinking in LLM-Supported Learning, explores an approach the researchers describe as "positive friction". Instead of giving students immediate and definitive answers, AI systems would introduce carefully designed ambiguity through open-ended or incomplete responses. The purpose is to encourage students to question information, evaluate alternative interpretations and, ultimately, develop their own positions.

 

A central outcome will be the development of ‘Diffrakt’, an AI-powered educational chatbot designed to support the teaching of AI ethics. It will take an approach of guided questioning and ambiguity with an objective to stimulate discussion and reflection rather than simply deliver ready-made answers.

 

The interdisciplinary research brings together expertise in human-computer interaction, artificial intelligence and learning sciences. It will initially be tested with students at ITU, with the longer-term ambition of informing educational technologies and practices beyond the university.


The project will be led by Professor Jichen Zhu, supported by Associate Professors Christian Hardmeier and Pawel Grabarczyk from ITU’s Center for Computing Education Research (CCER).


Why it matters for A4-AI (Academics for AI):


Much of the current discussion about AI in higher education focuses on concerns that students may become overly dependent on AI, potentially undermining the development of critical thinking and other traditional academic skills. There are also growing concerns about how students can continue to learn and be assessed with academic integrity in an environment where AI can increasingly perform tasks that were previously expected of them.

 

The ITU project introduces a different perspective to this discussion: what if AI could be designed to make students think harder, rather than simply make learning easier?

 

The concept of "positive friction" challenges the assumption that effective educational AI should always make learning faster and easier. Rather than positioning AI as a source of ready-made answers, it raises the possibility of AI becoming a tool that encourages students to question assumptions, consider alternative perspectives and develop their own judgement.

 

However, the effectiveness of this approach remains to be established.

 

The central assumption is that making AI responses less definitive will encourage students to think more critically. But the research will need to demonstrate that positive friction improves learning outcomes rather than merely making interactions more challenging.

 

There is also the challenge of students accepting the limitations of a chatbot such as the proposed Diffrakt rather than turning to the less restrained alternatives like ChatGPT.

 

But, perhaps, the most important challenge is how to demonstrate that students are actually becoming better critical thinkers. The research will need to establish meaningful ways of assessing improvements in independent reasoning, critical evaluation and the ability to apply knowledge beyond the AI interaction.

 

It will certainly be an interesting initiative to keep an eye on.


Questions for Academia:


  • Could introducing deliberate ambiguity into educational AI help address concerns about students becoming overly dependent on generative AI?


  • In practice, are students likely to adopt AI systems that deliberately withhold immediate answers when less restricted alternatives are readily available?





Comments


bottom of page