Home

On Learning Software Engineering in the Age of GenAI

Published

During the last couple of days, I have been attending Web Summit in Lisbon. Beyond my general impressions of this and other technology fairs, I had a couple of genuinely interesting conversations about the topic of the hour: AI and its effects on both the economy as a whole and software engineering as a discipline.

I have no intention of playing futurologist or looking into a crystal ball to answer those questions. However, one question in particular keeps arising with an alarmist tone attached. Following the “rule of three for blogging,” I decided to write this short piece so I could refer to it later.

The argument goes like this: in the world we live in today, there is little incentive to learn software engineering because most, if not all, tasks can now be reliably accomplished by a frontier model. The logical follow-up corollary is that teaching this discipline is no longer useful and that, if we stop teaching it, we will eventually reach a point where nobody understands these topics anymore.

Heavy AI use has been shown to atrophy our abilities, and this is painfully visible to anyone paying attention. Nevertheless, I consider this particular argument misguided because it assumes that there is no adaptation or alternative approach to consider when it comes to education.

The technology world changes quickly enough that new and more powerful abstractions appear semi-regularly. When I was studying computer science, we learned assembly language and even programmed simple algorithms in it. We then closed that section and moved on to more modern programming languages, such as Pascal and Fortran, which were soon replaced in our studies by object-oriented languages such as Java. Assembly is useful and interesting to know and understand, but it is not the environment in which we were ultimately expected to work. The reason is the abstraction layer on which much of the technology sits.

Today, AI plays a similar role by abstracting parts of the underlying stack. As a result, learning some of the nooks and crannies by heart is no longer necessary or even recommended. A broad understanding remains important and should not be abandoned, but being able to work with current tools efficiently and safely is itself a skill that must be learned and honed. The fact that instructions are written in natural language does not mean that they can be used productively without additional knowledge.

University and boot-camp training programs therefore need to adapt to this new way of working without necessarily discarding the existing syllabus, which remains relevant, if only in a more supporting role. This is a new way of approaching education that serves as a starting point for the job market, but it is also increasingly what recruiters look for and what is used in day-to-day work.

Disregarding this shift and listening to doomsayers who talk about the democratization or even the end of software engineering shows a lack of understanding of what many professionals in this field accomplish around the world. Tomorrow could be different, but for the moment, education should adapt, as the rest of the world does.