24 pilot courses, more than 1,000 students and four European higher education institutions. What happens when artificial intelligence is deliberately integrated into teaching and learning? The AI-HED project explored this question, generating valuable insights into how AI is changing students’ learning, educators’ roles, curricula and assessment practices.
At the AI-HED Final Event, held on 6 October 2026 at the Amsterdam University of Applied Sciences (AUAS), Erik Kostelijk, Associate Professor at AUAS and AI-HED project coordinator, presented the project’s main findings in a session entitled “What the AI-HED Pilots Teach Us”.
His presentation addressed a fundamental challenge: how can higher education move from individual experimentation with AI towards its effective, responsible and sustainable integration into teaching and learning?
AI-Enhanced Courses
Students
Teachers Would Highly Recommend AI Integration
From experimentation to evidence: the AI-HED pilots
Over two years, the Erasmus+ AI-HED project brought together four European higher education institutions from the Netherlands, Austria, Portugal and Croatia to explore the implications of AI for higher education.
Following initial activities focused on developing educators’ AI literacy and preparing pilot courses, the consortium implemented 24 AI-enhanced courses across different disciplines, including marketing, communication, journalism, languages and business process management.
More than 1,000 students participated in courses where AI was deliberately integrated into teaching and learning activities.
To understand the impact of these experiences, the project adopted a structured evaluation approach, combining pre- and post-semester surveys with teachers’ weekly logbooks and focus groups involving students and educators.
The findings revealed encouraging developments, alongside important questions about the future of higher education.
Students developed greater AI competence and confidence
One of the clearest findings was the improvement in students’ self-reported AI-related digital competencies across five dimensions: information and data literacy, communication and collaboration, digital content creation, safety and responsible use, and problem-solving.
Following the pilot semester, students reported higher competency levels across all five areas, particularly in information searching, content creation, communication and responsible AI use.
Students’ confidence in using AI also increased: the proportion reporting high confidence rose from approximately 40% to more than 50%, while positive attitudes towards using AI in a professional context increased from 33% to 45%.
However, the findings also revealed an important distinction between using AI and using it critically.
Students gradually moved beyond treating AI as a simple question-answering tool towards more sophisticated interactions, including questioning, refining and critically evaluating AI-generated responses.
This development highlighted the continued importance of disciplinary knowledge. Students need sufficient understanding of a subject to interpret AI outputs, identify weaknesses and make informed decisions.
AI competence and disciplinary competence must develop together.
Although responsible and safe AI use improved, it remained one of the areas with the lowest self-reported competency levels, reinforcing the need for continued attention to critical and responsible AI literacy.
Educators also learned — and their role is evolving
The impact of the pilots was not limited to students.
Educators reported substantial improvements in their AI-related competencies, with increases across all five digital competence dimensions. Their confidence in using AI for teaching also increased, from below 40% to approximately 60%.
Perhaps one of the most encouraging findings was that 70% of participating educators would highly recommend integrating AI into teaching to their colleagues, with none expressing opposition.
The pilots also highlighted how AI can reinforce the educator’s role as a facilitator of learning.
By supporting students with initial questions and routine tasks, AI created opportunities for teachers to focus more on deeper discussions, individual guidance, critical reflection and meaningful feedback.
Rather than diminishing the educator’s role, effective AI integration can make pedagogical expertise and human interaction even more valuable.
These findings also demonstrate that educators’ AI competence develops through practical experience, not only through theoretical training.
AI is changing what we teach — and how we assess learning
Using Bloom’s taxonomy as a reference, the AI-HED evaluation examined how AI supports different levels of cognitive learning.
The findings suggested that AI can contribute not only to basic learning activities but also to higher-order cognitive processes, particularly analysing, evaluating and creating.
Students and teachers reported benefits in understanding complex concepts, applying knowledge and supporting learning activities. More than 90% of participating teachers agreed that AI integration improved the overall learning experience in their courses.
However, these positive findings also raised important questions about curriculum design.
If students increasingly use AI to complete complex tasks, how can higher education ensure that they still develop the foundational knowledge and competencies required to evaluate those outputs?
Erik highlighted the need to integrate AI progressively throughout study programmes, aligning its use with students’ growing disciplinary knowledge and professional competence.
Assessment presents another fundamental challenge.
As AI becomes increasingly capable of producing high-quality written outputs, the final product alone can no longer provide sufficient evidence of what a student actually knows or can do.
The focus of assessment must therefore shift towards demonstrating competence, explaining decisions and establishing student ownership of the learning process.
This may require greater use of oral assessment, process-based evidence and opportunities for students to explain and defend their work.
However, these approaches also demand more time and resources from educators, creating a challenge for institutions seeking to scale AI integration beyond individual pilot courses.
The risks: learning, equity and institutional adaptability
Despite the positive outcomes, the project identified several risks that higher education institutions must address.
One of the most significant concerns is the difference between improving task performance and improving learning.
AI can help students produce better results, but it can also perform cognitive work that students need to develop independently. A high-quality assignment does not necessarily demonstrate that meaningful learning has taken place.
Another challenge concerns equity.
Students do not necessarily have equal access to AI tools, particularly when more advanced capabilities require paid subscriptions. If access to certain tools becomes necessary for learning, differences in financial resources may translate into unequal educational opportunities.
Finally, universities face the challenge of adapting to a technology that evolves much faster than traditional institutional processes.
Individual pilot courses can demonstrate what works, but sustainable integration requires coordinated approaches across programmes and institutions.
Five recommendations: from improvisation to effective integration
Drawing on the results of the pilots, Erik presented five recommendations for higher education institutions.
1. Rethink education, not just AI use
AI is changing how students learn. Universities must reconsider what students should learn, how learning takes place and how competence is recognised, rather than simply adding AI tools to existing educational models.
2. Build progression across the student journey
Individual courses can innovate, but coherent development of AI and disciplinary competencies requires programme-level coordination. Institutions should establish clear frameworks, programmes should design progressive learning pathways, and educators should develop learning activities aligned with these objectives.
3. Make competence visible without making assessment unmanageable
Assessment must provide evidence of students’ understanding, decision-making and ownership of their work. Institutions need approaches that combine meaningful assessment with realistic workloads for educators.
4. Ensure equitable access to AI
If AI is an essential part of learning, access cannot depend on students’ financial resources. Universities must ensure that all students have access to tools with the capabilities required to achieve their learning objectives.
5. Build universities that keep learning
AI will continue to evolve, and higher education institutions must evolve with it. This requires continuous professional development, opportunities for experimentation, the sharing of good practices and institutional frameworks that can adapt to changing educational needs.
The future of AI in higher education is a pedagogical and institutional challenge
The AI-HED pilots demonstrated that meaningful AI integration can strengthen students’ competencies, support educators’ professional development and create valuable learning opportunities.
At the same time, the findings show that successful integration requires more than technological proficiency. It demands pedagogical innovation, curriculum alignment, new approaches to assessment, equitable access and institutional adaptability.
Above all, AI should enhance learning without replacing the cognitive effort, critical judgement and responsibility that education aims to develop.
The central message from AI-HED is clear: moving from improvisation to effective integration requires universities that are willing and able to keep learning.
Watch the presentation and explore the AI-HED results
Interested in discovering the evidence, findings and recommendations behind these conclusions?
Watch Erik Kostelijk's presentation Download the slides (PDF)
Discover how educators translated these insights into classroom practices:
