If Everyone Is Using AI, What Does That Mean for Education?

A recent survey by the Digital Education Council found that 86% of students already use AI in their studies, and nearly a quarter rely on it every day. Those numbers are easy to read past. But sit with them for a second: AI isn't coming to education. It's already here, in almost every backpack, on almost every laptop, open in a tab next to the lecture slides.

This is a question we care about more than most. We're Arkintel, and our education product, JustAskMomo, has been running inside real lectures at TU Delft's Faculty of Aerospace Engineering. So instead of guessing what AI means for learning, we filmed the people living it: the students using it every day, and the professors rethinking how they teach. Here is what they told us.

"I don't know anyone who wasn't using it"

Walk through an engineering faculty today and the consensus is immediate. "I don't know anyone who wasn't using AI while studying," one student told us. "I think everyone does."

That ubiquity is the whole point. The interesting question isn't whether students use AI, because that debate is over. It's what happens to understanding, creativity, and assessment when a tool this powerful is always within reach. And the people closest to it, the ones teaching and learning every day, have surprisingly clear-eyed answers.

The goal was never the answer. It was the understanding.

For Joris Melkert, Associate Professor and Director of Education at the Faculty of Aerospace Engineering, the concern starts with what education is for.

"We don't want people to memorize formulas. We want them to understand formulas, to understand what they're doing in design, what they're doing in analysis."

His worry isn't that students cheat. It's subtler than that. AI is built on past performance: it makes links between things that have already been done. So the question that keeps him up is whether the next generation stays creative enough to take the step that hasn't been done yet.

"AI makes links, but are those links creative, or do you still need the human being? I hope you still need the human being. And we need to train that. We also need the critical attitudes to be trained."

"Saying you can never use AI is stupid"

Calvin Rans, Associate Professor in Aerospace Structures and Materials and on an education-focused track, comes at it from the other direction. Not what to forbid, but where to draw the line.

"If we just clear-cut say you can never use AI, well, that's stupid. You're going to have to use it in your career at some point. You need to get familiar with it."

His analogy is the calculator. AI is a tool like any other, and the skill is knowing when to reach for it.

"If I'm teaching you integration and you use a graphing calculator to do the numerical integration for you all the time, then you don't learn about integration. You have to know the appropriate use of a tool in a given situation."

That's the line every student is now learning to walk as they figure out where to draw it.

The students already know the trade-off

Jung Kyu Kim, a first-year master's student in Aerospace Structures and Materials, has a rule for himself.

"I think it should be a tool that helps on the side, not one that does everything for you."

Especially with code. "When it comes to coding, it's really important that you struggle for a couple of hours," he says. The frustration is the learning.

But that struggle is changing. Kyra Valk, Commissioner of Master Affairs on the VSV student board, has watched her own relationship with learning shift in real time.

"I'm more reliant on it than I would have been without it. When I did Python in my first year, AI wasn't really a thing yet, so I did it all on my own. Now the Python course would look a whole lot different. You'd lean on AI more, and maybe you wouldn't learn it the way you would have if it weren't there."

What she's naming is something many students feel: AI makes things easier, and "easier" isn't always the same as "learned."

Assessment is quietly breaking

If learning is changing, the way we measure it has to change with it. Here the professors are blunt.

"A take-home exam, a written report, a take-home assignment has become worthless as a tool of assessment," Melkert says. "You can no longer tell whether the student did it themselves, did it with help from AI, or had it fully written by AI."

His answer is to add a layer on top: if a student hands in a report, you now have to check whether the knowledge in the report is also in their head.

Rans points to a deeper structural problem. Much of higher education runs on a kind of factory model, with huge cohorts and a classical exam that asks a question with one defined output.

"I give you a question that has a very defined output, and I check if you get that number. That's a problem in an AI system, because AI helps you get the number. It doesn't necessarily test your conceptual understanding."

His conclusion is almost paradoxical: the fix for AI-disrupted assessment may be AI itself.

"Ironically, what will eventually have to happen is that we start using AI to help us assess. Not doing everything, but helping."

Where AI earns its place

For all the caution, no one we spoke to wanted AI gone. The value is real and specific.

Valk uses it as a translator for hard ideas. A textbook explains a concept one way, a lecturer explains it another, and AI gives her a third angle.

"AI can explain what you need to learn in a different way than a teacher does, or than your book does. You get a different point of view on the material, and that helps, especially in courses that were too abstract for me."

Melkert sees the same upside at the professional end. Engineers increasingly end up writing reports and consultancy decks instead of doing engineering.

"Now you ask AI to write the consultancy report, so engineers can do more engineering. I see very positive sides to this."

The tools, as he puts it, are here to stay: useful in learning, in decision-making, in data analysis. The job isn't to resist them. It's to find the right way to work with them.

The gap generic AI can't close

There's one limitation every student named: generic AI doesn't know your course. Ask a general chatbot a subject-specific question and you get plausible, confident, sometimes-wrong answers, drawn from the entire internet rather than the material you're actually being examined on.

That gap is the reason we built JustAskMomo. We made it in Delft and tested it inside real aerospace lectures: an AI study companion grounded in the actual course content, so its answers line up with what's taught rather than with whatever the open web happens to contain.

Jung Kyu Kim felt the difference immediately.

"With a specific subject like this, where you need that scientific knowledge, a general chatbot just gives me random things. JustAskMomo was only giving me things from the course itself. That I really found helpful."

For dense material, it lowered the barrier to even getting started.

"There's so much to read, and half the time I have no idea what the text is saying. I ask what it means, it summarizes it, and then I can read it and go, oh, that makes more sense."

The payoff that matters most to educators is this: when the AI is anchored to the course, students don't quietly absorb wrong information. As one student put it, "if it has course-specific information, you don't learn anything wrong from AI."

From raw data to wisdom

Melkert has a way of describing the arc of education that stuck with us.

"It always starts with raw data. In the end, you want to create wisdom."

The teacher's role, he says, is to make the links, to bring the knowledge but also show the connections, so that students develop the deeper understanding that eventually becomes wisdom. AI can help along that path. How far it goes, he admits, "the future will tell. It's still hidden in the stars."

That's the crossroads education is standing at. AI is now an everyday tool, and pretending otherwise helps no one. The real work is to nurture the things only humans bring, like creativity, critical thinking, and ethical judgment, while letting AI do what it's good at: processing information and offering a different way in.

That balance is exactly what we're building toward. JustAskMomo is where Arkintel began — a course-grounded study companion built in Delft with one conviction: that AI should respect boundaries. That same conviction now runs through the full Arkintel office suite: chat, docs, drive, sheets, slides and more, AI-native and EU-hosted, with your data governed by European law and never used to train a model.

AI in education isn't a threat to manage or a shortcut to ban. It's a tool to teach well. The students already know that. So do their professors. The task now is to build the tools that live up to it.

JustAskMomo was where it started. See what it grew into — the Arkintel office suite. Universities and governments can see how we deploy in production.