Are we rewriting the curriculum or are we piling up tools?
This week, a systematic review was published that I couldn’t get out of my mind. Sabani and colleagues (2026) analysed 209 studies on generative AI in higher education and mapped out what all that research is actually about.
Their most striking finding lies in what is missing. In their bibliometric map, the curriculum appears as a peripheral concept. Almost all attention is focused on tools, assessment, and fraud. The place where education is truly designed remains largely out of sight.
We recognise this. As a lecturer, Pascal Mariany sees enthusiastic experiments everywhere. As a founder, he speaks to administrators struggling with policy. But the conversation about the curriculum as a whole is rarely held.
Bakogianni, Liljekvist, and Bui (2026) therefore call generative AI a runaway object. Adoption takes place via students and colleagues, informally and outside the institution. Meanwhile, curriculum committees are waiting on the sidelines.
Five Shifts
Sabani and colleagues describe five shifts that together form a new learning system.
From a static curriculum to dynamic design in which AI is both content and method
From teacher-led transmission to guidance in which AI collaborates
From knowledge transfer to developing sustainable capabilities
From isolated experiments to institutional agreements
From fragmented tools to a coherent ecosystem
Their first proposition sums it up. Integration of AI requires rethinking the curriculum, not just replacing individual components.
Why This Is Urgent
Other recent research shows what happens if we do not have this conversation.
Crolla, Xia, and Jiang (2026) found that AI increases the differences between students. Strong students use AI to expand their thinking. Vulnerable students outsource the thinking and submit polished products without understanding.
Chang and Li (2026) analysed over sixty thousand student conversations with an AI assistant. It was not the individual student but the design of the course that determined how deeply students thought.
Wang and colleagues (2026) aptly describe the risk as performance without learning. And Mathew and colleagues (2026) show that language models assess essays fundamentally differently from humans. The professional judgement of the teacher remains indispensable.
Deng, Çelik, and Duran (2026) also conclude that teachers’ AI literacy currently arises mainly through individual pioneering. Sustainable development requires institutional structures.
Each of these studies points in the same direction. The question is not which tool we choose, but how we design learning.
Looking with HEAR
In the book The Art of Living Together with AI (Mariany, 2026), Pascal introduces the HEAR framework for human-centred work with AI. Four layers together determine whether the interaction with AI is educational, responsible, and formative.
Human Intent. The intent comes first. What do we want students to develop here, and under what conditions?
Expressive Interaction. The language between human and system, which can sound convincing even though the system means nothing.
AI Mediation. What happens between question and answer. AI selects, organises, and narrows thinking paths even before we make a judgement.
Reasoned Responsibility. Ownership of the judgement, grounded in traceable considerations.
At curriculum level, HEAR gains an extra dimension. Human Intent then no longer lies with a single teacher but with the team. And Reasoned Responsibility becomes collective. Not one teacher, but the programme weighs, decides, and legitimises.
This is precisely the shift Sabani and colleagues describe as the step from isolated experiments to a coherent learning system.
Curriculum Thinking in Practice
Especially in an era of AI, we believe that teachers are designers, not mere implementers of a tool. With that conviction, we built the Curriculum Companion within EduGPT.
The companion does not rewrite a curriculum or make judgements. It poses design questions. Where is independence assumed without being prepared for? Where is feedback structurally embedded in the learning pathway, and where only incidentally? Where does AI add value, and where should AI be consciously absent?
Teams working with it notice that the conversation shifts. From isolated tools and pilots to design principles. From “what must we implement” to “what do we consider important enough to choose explicitly”.
The thinking remains where it belongs. With people.
If you want to see this in action, you’ll find knowledge clips about the companions on our YouTube channel. In the EduGPT demo environment, you can try out the Curriculum Companion yourself with your own curriculum document.
Never Done Learning
The research by Sabani and colleagues confirms what we see every day. The technology is there and the enthusiasm is there. The real challenge lies in design and coherence.
We would be happy to discuss this further. Is your programme still discussing AI at the level of tools, or already at the level of the curriculum? We are curious to know where you stand.
Sources
Bakogianni, D., Liljekvist, Y., & Bui, P. (2026). GenAI as a runaway object in higher education: A socio-cultural view on AI-influenced academic practice in mathematics. Computers and Education: Artificial Intelligence, 11, Article 100637. https://doi.org/10.1016/j.caeai.2026.100637
Chang, C.-K., & Li, K.-H. (2026). Chat as learning: Student–AI conversations as discipline-associated cognitive engagement patterns. Computers and Education: Artificial Intelligence, 11, Article 100644. https://doi.org/10.1016/j.caeai.2026.100644
Crolla, K., Xia, X., & Jiang, Y. (2026). AI-mediated cognitive divergence in built-environment education: Evidence from a mixed-methods study. Computers and Education: Artificial Intelligence, 11, Article 100665. https://doi.org/10.1016/j.caeai.2026.100665
Deng, Y., Çelik, F., & Duran, V. (2026). Governing the unseen: A systematic review of AI literacy among language teachers in higher education. Computers and Education: Artificial Intelligence, 11, Article 100658. https://doi.org/10.1016/j.caeai.2026.100658
Mariany, P. (2026). The Art of Living Together with AI. Van Duuren Media.
Mathew, J. G., Taher, S., Kundu, A., & Barbosa, D. (2026). LLMs do not grade essays like humans. Computers and Education: Artificial Intelligence, 11, Article 100666. https://doi.org/10.1016/j.caeai.2026.100666
Sabani, A., Farah, M. H., Catyanadika, P. E., Dewi, D. R. S., & Tawani, V. (2026). Rewriting the curriculum: A systematic review of generative AI-driven pedagogical change and emerging systems of learning in higher education. Computers and Education: Artificial Intelligence, 11, Article 100667. https://doi.org/10.1016/j.caeai.2026.100667
Wang, X., Zheng, Z., Zhang, J., Hou, X., & Zhu, Z. (2026). The IDEA framework for metacognitively regulated GenAI use in higher education: Development and exploratory pilot evidence. Computers and Education: Artificial Intelligence, 11, Article 100657. https://doi.org/10.1016/j.caeai.2026.100657