I had the opportunity to participate for three days at the EDEN Conference 2026 in Porto. The theme “Beyond Technology: Human-AI Collaboration for Learning and Teaching” initially seemed more broad than deep. However, as it turned out, media-didactic and higher education practitioners and researchers navigate reflectively and carefully in an educational landscape that appears to be characterized primarily by uncertainty and pressure to act.
What I took away is:
- How the community uses AI – without following uncritically
- That instructors are increasingly becoming scenario architects in sociotechnical learning architectures
- That AI research (still) can contribute little to solving the open questions on “AI and education”
- Which didactic translations appear in AI-supported teaching formats
- How leadership and governance deal with uncertainty
My personal conclusion is that uncertainty is not a hindrance for a strong community.
Critical Affirmation
The tension field in which the AI-induced transformation in education was discussed at the EDEN conference unfolded between the concepts of continuity, disruption, urgency, and caution. In personal conversations, much insecurity and confusion were palpable; partly, this was also openly addressed. “We don’t know what we don’t know” was a central claim. On the other hand, the question whether education experts should adopt AI seems to be answered with “absolutely.” The question of meaning is still an academic (“philosophical”) sub-question, but no longer relevant to action.
What remains is astonishment that instructional designers, educational designers, and educational technology visionaries, who long understood themselves as pioneers of educational innovation, now find themselves in the role of warners, brakes, and skeptics. They appear today as appendages of corporate decisions, driven by market forces and geopolitical dynamics. I readily want to avoid adopting the pretentious “I-told-you-so” attitude, yet: the bitter awakening is also a consequence of perhaps too-neglected treatment of fundamental questions regarding the material foundations of one’s own discipline. What is the relationship between technology and education? What role does technology development play for us? Who determines this? How do we position ourselves regarding this?
That there is a critical-analytical need here was shown by the fact that the call to “Follow the Money” by one participant received both spontaneous applause and numerous references later in the conference. It seems to me that the community is only waiting for the overdue critical questions to be formulated. However, the practical and theoretical contributions of the conference also showed (in most cases…) that practical engagement and critical reflection are not contradictory. In almost all contributions, ways were sought and described to combine didactic action and design capacity with the principles of “human-centered,” inclusive, and critical AI use.
It is possible to adopt and use AI for teaching and instruction without proceeding uncritically and blindly to contradictions. The conference showed me that there are many approaches to critically affirming AI.
Hybrid Extension
An exciting development trend that apparently gains significance with AI use in instruction and teaching is the shift from “designing the course” or “didactic scenario” toward //sociotechnical learning architectures//. A further, more pronounced role shift for instructors, who are now supposed to become scenario architects. This is still quite vague and unclear, but the planned use of “communicating” technologies and “autonomous” agents provokes splitting the instructor role and distributing it onto the technology. Scenario creation, contextualization, and design of technological settings for learning become the topic.
That simple role transfers (“AI as learning partner”) quickly reach theoretical limits, I was able to put up for discussion together with Ulrike Schroer in our own contribution1. To advance conceptually there, it would be sensible to examine the relationship between education and technology more closely. So far, I see little new here2. In any case, overly simple constructs of technology, such as the metaphor of the //tool//, are now at risk.
Interweaving and mixing, entangling and weaving, from my perspective are good linguistic markers indicating the direction in which one can think. This concerns competencies, responsibility, cognition, and other core concepts of pedagogy, learning theory, and didactics. The hybridity of the objects that together make up what we call our “world” seems to me an analytic perspective that best expresses this ambiguity3. Educational Design (“Didaktik” in German) as a discipline dealing with “theory and practice of teaching and learning” does well to continue critically questioning its conceptual boundaries and distinctions. The practice of research and theory-guided practices flow into each other; instructors must learn, learners must formulate what they expect from teaching and their learning. Overcoming the dichotomies between pedagogy and technology, education and technology seem to me the challenges that arise in an “AI-shaped future” for educational institutions and individuals.
Benefit of AI Research
What the EDEN conference, as far as I can judge, could not achieve was to provide an outlook on what AI research can or wants to contribute to solving the open questions on “AI and education.” The insights I received showed little more than that they once again clearly demonstrated that a complex structure of hardware, data, operating systems, machine-learning (ML) frameworks, and others produces what we experience as “AI.” This once again made clear to me that “educational technology” is something fundamentally different than “technology used in education.” From the AI researcher’s perspective, the message was: “If you know what you want to do, then you can also use AI.” This sounds to me like an attempt to reduce AI to a easily graspable “tool model.” That may be sufficient for development, but it is not for application.
For example, I would like to ask AI research the following: Is using AI a zero-sum game? Solutions arise by shifting problems toward other unsolved and multiplying problems. I might get a train schedule inquiry, feedback on my text, or a draft for a new molecule. But for this, an unknown (though certainly large) number of natural and human resources are depleted and used. The “magic” (the apparent added value) of the AI output essentially rests on the fact that this whole ensemble behind it is not visible to me. What does the problem-solving balance look like? A small delicate paradox: Exactly the presentation on the state of AI research was a good example of something that could well be replaced by an LLM without significant losses, “… and we have a lot of tools…”.
Ed-Design Inscriptions
In many contributions I was able to attend, interventions were reported on how AI is used and evaluated in teaching and instruction. It can be noted that here too, and especially with AI use, the well-known media-didactic principle shows that using a tool, resource, or medium cannot be understood as simply swapping one element while everything else remains constant. First, AI use usually affects all other elements; second, this use comes belatedly for students, meaning they already use AI for everything; and third, scenarios are typically “built around the AI” (see above the “scenario architects”).
The scenarios vary: Sometimes the LLM is used deliberately for formative feedback, whether this already “makes students’ thinking visible” was unclear to me. Others use AI-supported preparation of students for an exam format in the form of a case-based expert interview. The experience here: students use it more for analyzing the case than for practicing the interview. With complex questions about the case, the LLM apparently could not cope. It was investigated, among other things, whether using AI has effects on students that can be reconciled with Cognitive Load Theory (CLT). From the results, a critical assessment of the benefit of AI emerged. This concerns at least more complex learning outcomes such as transferring what was learned to new situations. From this it was also derived that CLT would need to be expanded by including AI. In the area of teacher continuing education, rather classic media-pedagogical interventions were presented. They revolve around the questions: What are AI tools? How can I use them? How can I design them? Here too, the clear commitment was central that using AI is particularly a “social challenge.” It would certainly be interesting to follow how “sociality” is constructed in these contexts. Values? Taboos? Power structures? Institutions?
Leading in Uncertain Mode
The topics Leadership and Governance play a large role at EDEN, which shows that participants and active individuals are partly involved in leadership positions and strategic action fields. I found instructive a jointly developed statement in which, from the perspective of leaders, it was recorded what the desired state regarding AI treatment at the university could be:
- From crisis to control of the situation.
- Strengthen democracy and prevent digital division.
- Promote responsible use and critical thinking.
- Transparency of origin and fair sharing of content.
- Preserve originality and independence.
- Involve all stakeholder groups.
- Clarify guidelines and frameworks.
Orientation for leaders and strategic suggestions were offered in some contributions. These usually referred to general analyses of the action field and the definition of dimensions for action. For me, this included that study success can be defined and structurally framed (even if this does not immediately translate into metrics) or maturity models for digitizing teaching. The contribution from HIS should also be mentioned here, which provides a solid empirical basis on the state of AI use at German universities. A bridge was struck by the experience report from the USA, which showed how the discussion about a framework for AI literacy is particularly well suited to initiate university-wide engagement with AI and discussion.
Overall, even in the topics of leadership and strategic considerations, according to my perception, the handling of uncertainty and a crisis experience associated with AI is dominant. A obvious recommendation is therefore to rely on cross-functional teams and work structures to achieve a new setup affecting all areas of the university. From my perspective, it would be an interesting desideratum to learn more about what actually happens in such teams and work groups. What input ends there, how is it processed? What action and influence possibilities does the group have, do individual participants have? Where is there agreement, where do contradictions appear? Will we find results, record them, and pass them on? Presumably, the internal dynamics in such teams give clear indications how obstacles and opportunities for AI at the university shape and distribute.
Conclusion: Uncertainty Is Not a Hindrance for a Strong Community
My personal conclusion from EDEN 2026 is that uncertainty regarding AI in higher education does not prevent implementing a reflective approach committed to pedagogical progress. The many question marks we still have today are perceived as challenges that are often met with proven means. These include at EDEN curiosity, desire to design, and trust in a strong community.
I also want to point to the very informative contribution by Klaus Wannemacher (HIS) on the EDEN conference (in German).
(translated with massive help of Perplexity)
- Ulrike Schroer, Jörg Hafer: Beyond the Machine: A Proposal for an Actor Network and Practice-Theoretical Perspective on GenAI in Higher Education (Abstract) [↩]
- One approach that makes the nature of AI quite tangible is Werner Rammert’s concept of “autonomous machines” [↩]
- Bruno Latour: We Have Never Been Modern [↩]
