In fact, we need more IT professionals (column: Erdinç Saçan)

With AI, you can now build an app in an afternoon. Just enter a few prompts, click a few times and publish. In other words: vibe coding. You describe what you want, the AI writes the code, and before you know it, you’ve got something working on your screen.

Don’t get me wrong: I think that’s brilliant. Want to create a script that automatically clears out your inbox? Fine. Want to build a prototype to test an idea or show it to colleagues? Even better.

But a recent incident in Switzerland shows where things can go wrong. A medical professional used AI to build their own patient management system. Patient data was imported, consultations were routed to AI services, and the system was put online. It then transpired that a security specialist was able to gain access to all patient records within a short space of time.

The system worked. That was precisely the problem. Because it worked, it was used.

Romano Roth put it aptly on LinkedIn: this is not the democratisation of software, but the democratisation of liability. As soon as software processes real personal data, the story doesn’t end with a working application. Then it’s about security, privacy, legislation, access rights, management and responsibility.

This ties in nicely with a recent article by researchers Arvind Narayanan and Sayash Kapoor. They describe software development as much more than just writing code. First, you have to decide what you’re going to build. Then comes the implementation. Next, you have to check whether it can be used safely, reliably and responsibly. It is precisely those first and last steps that often turn out to be the most labour-intensive.

I see this reflected in education too. These days, students can create impressive things in a short space of time – a chatbot, a dashboard or even a complete application. But in conversations with students, I notice that the most interesting questions are increasingly less about code. Where is data stored? Who has access? What risks are inherent in the system? What happens if an AI model makes the wrong decision? And who is responsible for that?

Perhaps that is the biggest misconception at the moment: that AI would mean we need fewer IT professionals. Yet the opposite is happening. The easier it becomes to build software, the more important it is to have people who understand how systems work, recognise risks and can assess the consequences of decisions. But IT education must adapt to this too! As soon as possible.

Erdinç Saçan teaches at Fontys University of Applied Sciences’ ICT department and is an AI expert. Last year, he published the book AI for Everyone. Read here his previous columns for Bron.