The Invisible Assistant

For centuries, studying meant working on practice sheets, poring over academic articles in the library, or staring at a blank screen while writing a term paper. But ever since late 2022, almost every student’s laptop has come equipped with a tool that can do all of this in mere seconds: It writes texts, explains evidence, writes programs, and summarizes literature. Overnight, artificial intelligence has become an integral part of teaching, learning, and testing—and it is prompting us to rethink what it actually means to be a student. At RWTH, we are actively shaping this change rather than merely managing its effects.

Learning Around the Clock

For today’s students, AI has already become a normal part of everyday life. Students can have thermodynamics concepts explained over and over again until they understand them, they can practice exam questions and receive immediate feedback, and they can even debug code in the middle of the night when no tutors are available. Having someone explain something to you patiently at exactly the right moment used to be a scarce resource, limited by office hours and staff capacity—but today, it’s available right around the clock.

But this invisible assistant does come with an important drawback. Having a solution handed to you instead of working it out for yourself often leads to you learning less than you think you have. Experts refer to this as “deskilling”—gradually losing skills that you delegate to others rather than practicing them yourself. A discussion paper by the Higher Education Forum on Digitalization rightly asks whether the widespread euphoria surrounding new “future skills” might be obscuring our view of the skills that we could lose in the process. What’s more: AI systems sound confident, even when they’re wrong—and first-year college students often lack the knowledge to recognize plausible misinformation. The focus of learning is shifting: It’s no longer just “Can I work this out?” but rather “Can I tell if the result that has been given to me is correct?”

Education That Is Reinventing Itself

A lot is changing on the other side of the lecture hall, too. Instructors use AI to create materials more quickly, vary practice exercises, or provide students with personalized feedback—at a level of quality and quantity that would never be feasible manually in lecture halls with several hundred students.

On the other hand, there is a more fundamental question that we should answer: What skills do we actually want to teach? Which subject-specific knowledge and interdisciplinary skills should remain firmly rooted in our students’ minds—and which ones can we do without? In the AI debate, people often draw a comparison with the calculator, which is said to have made basic math skills obsolete. But this comparison doesn't really hold water: Even in 2026, our children are still learning how to do written division. If we apply the same theory to AI, this would mean that outsourcing thinking and writing to AI on a permanent basis would be a disastrous development. The real challenge is this: How can we ensure that students don’t take the easy way out when they’re sitting alone at their desks facing a difficult task, and a quick AI fix is just one prompt away?

RWTH has an important message to convey here: AI cannot replace teachers. It alters their role. Any system today can retrieve pure factual knowledge; the added value of good teaching lies in contextualizing, questioning, and guiding students. That is why RWTH is also participating in projects at the state and federal levels, such as AI:edu.nrw, and, in the future, the AI Expertise Center, to maintain a continuous balance between technical feasibility and educational value.

Exams and Assessments are of Critical Importance Here

Exams and assessments are when this matter becomes most prominent. The traditional term paper as a measure of academic performance can no longer function as it once did when a language model can generate large portions of it in a matter of minutes. This is compounded by the fact that these AI tools can deliver such work with increasingly high levels of quality. This is not a minor issue, rather it is a fundamental question for academia: What exactly should we award points for?

Bans are not the right answer here—they are virtually impossible to enforce and they fail to take into account the realities of the workplace. Our approach focuses on better exams and new exam formats rather than on monitoring. We employ skill and process-oriented formats that show how someone arrives at a result, we apply oral and application-based exams, and transparent rules that demonstrate where AI is permitted and where it is not. Last but not least: learning how to use AI becomes a bona fide learning objective itself. Students today should be able to use these tools confidently, critically, and honestly.

What the University Must Accomplish

Such a strong creative drive does have a sobering downside. A public university cannot simply direct its students to private chatbot accounts—data protection and the requirements of the GDPR set clear requirements for this. We need secure, privacy-compliant, and functional solutions, and procuring them is significantly more time-consuming for a public institution than signing up for a quick subscription. RWTH has developed its own platform, RWTHgpt. This is a spin-off of the state-level project that is supported by RWTH. KI:connect.nrw. We view this as being an important first step in the right direction. The project KI:Inferenz.nrw, runs its own AI models and it makes us able to work independently of commercial international AI providers. As such, RWTH is taking a major step in the right direction here as well. In addition to this, teacher training is a core mission of the Center for Teaching and Learning Services (CLS), because even the best technology is of little use if there is no pedagogical method in place.

A Look Ahead

AI proficiency is no longer just a matter that is relevant to academic study, rather it is a key skill in the workplace. Many of our alumni are currently experiencing the same period of upheaval in their own organizations that we are going through here at the University. If we do it right, our graduates will leave RWTH not as users who blindly trust a machine—but as experts who know when they can trust it and when they cannot. At its core, that has always been the goal of a good academic education. The tools may have changed, but the goal remains the same.

– Author: Univ.-Prof. Dr. Malte Persike, Academic Director of the Center for Teaching and Learning Services (CLS) at RWTH Aachen University