AI in higher education

If you have the privilege of teaching in an IT program, change is part of your job. The way those changes looked over the past twenty years was neatly dictated by the industry; we just had to follow along. Today is different. The disruption caused by AI is so massive that there does not yet seem to be any clarity on where we should be heading. As a result, we see many well-intentioned initiatives shooting off in all directions. Suddenly, AI experts are also popping up everywhere—a label that seems very cheap these days. In any case, I am certainly not one of them, to be clear.

The trilemma

AI is a disruptive technology that is already having a huge impact on the IT industry. There is no need to waste more words on that; nobody needs to be convinced of it.

But how education should deal with it is much less clear to me. Specifically for IT skills like programming, the question quickly bubbles to the surface of what knowledge is still necessary, and at what point AI can take over. At one extreme of the spectrum are colleagues who believe you can only program well with AI if you are a good programmer yourself; at the other extreme, programming knowledge has become obsolete and tools are now leading, rather than strict knowledge. I’ll leave what it will be to the experts, and undoubtedly the truth—as so often—lies somewhere in the middle.

Either way, it is getting more and more attention in the curriculum, and that brings new choices to light. For instance, one observation is that AI threatens to quickly lead to higher pass rates for those who have the money to buy the most expensive and best models on their own initiative. This becomes apparent daily in the lab where students working with free tools sit shoulder to shoulder with students who sometimes even have multiple subscriptions with the likes of OpenAI and Anthropic.

Towards the next academic year, the decision will therefore likely be made to require students to provide identical AI tooling, so that everyone fights for a degree on equal terms. The textbook is out, the AI subscription is in.

The fascinating thing is that there are multiple ways to address this. It is by no means certain that the debate will be grounded in substance—technical discussions are more often than not decided on non-technical arguments—but it is at least worth a thought experiment. If we as an educational institution want to provide AI tooling, what is the preferred way to do it?

The three options:

• Option 1: let’s go shopping: We have all students buy a Claude subscription. That way we give them the tools that are on par with what is actually used in the industry. That is an extra cost for students, but IT studies are already very cheap compared to other fields anyway.

• Option 2: let’s go shopping in the EU: You don’t learn to drive in a Porsche either, and education has a moral obligation to give European alternatives room to grow and build a user base, even if statistics show those models lag about a year behind the American alternatives. By having students buy a Mistral subscription, we send a clear signal and provide a push for a stronger European AI footprint.

• Option 3: let’s build: A university of applied sciences is primarily a location where practice-oriented research and development should find a home. Instead of subscriptions for students (which amounts to roughly EUR 160,000 annually for 600 students¹), it would be better to invest in in-house AI servers that can be used by students and also benefit research. The experience gained by hosting models ourselves can then in turn be shared with the industry.

For this blog, I’m currently still a bit too lazy to provide interaction features, but I’ll turn it into a poll on LinkedIn and post the results here afterwards…

Footnotes

1. Source: Claude pro x 12 months x EUR20/month ↩