The Classroom Without Walls
How AI Can Make Students More Productive — and More Creative
Reflections inspired by Prof. Dr. Christoph Meinel, founding president of the German University of Digital Science
eyesoneurope
Amsterdam, July 17, 2026 – There is a particular kind of institution that only becomes possible once you stop assuming a lecture hall is required. The German University of Digital Science (German UDS), co-founded in 2025 by Christoph Meinel and Mike Friedrichsen, has no campus, no auditoriums, no chalk dust. Its founding motto — “University as a Service” — sounds at first like a piece of corporate jargon. But behind it sits a genuinely radical idea: that a university’s real product was never the building. It was always the access to thinking, delivered wherever a learner happens to be standing.
Meinel arrives at this vision with unusual credibility. Long before “generative AI” was a phrase anyone used at dinner parties, he spent nearly two decades running the Hasso Plattner Institute and building openHPI, one of Europe’s first MOOC platforms — teaching programming and computer science to hundreds of thousands of people who would never set foot in Potsdam. He has watched, in other words, what happens when you take education apart and rebuild it around the technology of the moment, not once but twice: first with the internet, now with AI.
That history matters, because it suggests the right question isn’t “should universities adopt AI?” That ship has sailed — students are already using it, whether or not their professors have caught up. The right question is the one German UDS was built to answer: what does a university look like when it takes AI seriously as infrastructure, not as a novelty bolted onto an old model?
Here is what that looks like in practice — and why it should make students, not just administrators, genuinely excited.
1. Teach AI as a collaborator, not a vending machine
The laziest way to bring AI into a classroom is to ban it. The second laziest way is to let it quietly become a homework-completing machine that students use in secret and professors pretend not to notice. Neither approach teaches anything.
The better path — one that shows up across German UDS’s Applied AI and Digital Leadership programs — is to make working with AI itself the skill being taught. That means assignments where a chatbot is a visible, acknowledged partner: a brainstorming companion during ideation, a devil’s advocate that pressure-tests an argument, a first-draft generator whose output the student is then required to interrogate, correct, and improve. The point isn’t to remove the struggle from learning. It’s to relocate it — from the mechanical parts of the work to the parts that actually build judgment: knowing what to ask, recognizing a wrong answer, deciding what’s worth keeping.
Best practice: design assignments around a visible AI-use trail. Ask students not just for a final answer but for the sequence of prompts, dead ends, and corrections that got them there. Grading the process, not just the product, is the single biggest lever a university has for turning AI from a cheating risk into a thinking amplifier.
2. Move assessment from the exam hall to the ongoing record
If a model can produce a competent essay in ten seconds, the traditional take-home essay stops measuring what it used to measure. Universities that cling to it are grading how well students can prompt, not how well they can think.
The more durable answer — one increasingly discussed across higher-education circles and central to how digitally-native institutions like German UDS structure their online, competency-based programs — is continuous, process-based assessment: oral defenses of written work, project logs, iterative portfolios, in-person or live-video checkpoints where a student has to explain and extend their own reasoning on the spot. This is harder to scale than a multiple-choice exam. It is also the only format AI genuinely cannot fake, because it tests something AI doesn’t have: a mind that has to stand behind its own conclusions.
Best practice: replace at least one high-stakes, one-shot exam per course with a running body of evidence — drafts, revisions, short live check-ins — that shows the trajectory of a student’s understanding, not just its final snapshot.
3. Design for global reach, not just local convenience
A quieter but equally important part of the German UDS model is who it’s built for. A fully digital, English-language university with no campus overhead can reach students in places a traditional institution never could — including, explicitly, learners across the Global South who would otherwise have no access to a Master’s in Applied AI or Cybersecurity at all.
This reframes what “productivity” means at the level of a whole system, not just an individual student. AI-native, low-overhead universities can offer modular micro-degrees, stackable credentials, and flexible course pacing that meet working adults and non-traditional students where they are — rather than demanding they rearrange their lives around a fixed semester and a fixed location.
Best practice: unbundle the degree. Offer credit-bearing micro-credentials that working professionals can stack over time, using AI-assisted, self-paced platforms to keep quality high without requiring physical attendance.
4. Protect — and deliberately train — the things AI can’t do for you
The most important thing an AI-forward university can do is also the most counterintuitive: spend more, not less, deliberate effort on distinctly human capacities. Design thinking, cross-cultural collaboration, ethical reasoning, the ability to sit with an ambiguous, half-formed problem before rushing to a solution — these don’t get automated away by better AI. If anything, they become more valuable, because they’re the scarce ingredient once the mechanical parts of research and drafting are cheap.
This is precisely why German UDS has invested in things like a dedicated College of Design Thinking, built around cross-cultural, integrative frameworks for tackling ambiguous problems — the kind of curriculum that treats creativity and judgment as core disciplines, not soft add-ons squeezed in around “real” coursework.
Best practice: build at least one required course per degree that has nothing to do with AI literacy and everything to do with ambiguity, ethics, and collaborative problem-framing — the terrain where human judgment still has no substitute.
5. Give faculty the same productivity gift you’re giving students
It’s easy to focus AI strategy entirely on students and forget the people designing their education. But an AI-native university treats faculty workload the same way it treats student workload: as something worth compressing, so the saved time can be reinvested in higher-value work. Automated first-pass grading of code and structured assignments, AI-assisted feedback drafting, and AI-supported course design all free instructors from repetitive work — not to replace their judgment, but to protect their bandwidth for mentorship, live discussion, and the kind of individualized guidance that no dashboard can generate.
Best practice: audit faculty time the same way you’d audit student time. Wherever a task is repetitive and rule-bound, let AI take the first pass — and reinvest the recovered hours in direct human contact.
The real shift
What connects all five of these practices isn’t a specific tool or platform. It’s a change in what a university believes it is for. If a university’s job was ever to be the sole gatekeeper of information, that job is already gone — a search bar took it, and then a chatbot took it more completely. What’s left, and what actually matters now, is the harder and more human work: teaching people to ask better questions, to hold an idea up to scrutiny, to build things worth building, and to know the difference between an answer that sounds right and one that is right.
That is a more ambitious mission than the one universities have quietly settled for over the last century. It is also, if institutions like German UDS are any indication, an entirely achievable one — not despite AI, but because of what AI frees us to focus on instead.
The lecture hall was never the point. The thinking was. AI, used well, is the first technology in a generation with the power to put that thinking back at the center of the university — for every student, wherever they happen to be standing.
