This opinion piece by the University of Waterloo’s Dr. Mary Wells, dean of engineering and Dr. Jochen Koenemann, dean of mathematics, appeared in the Hill Times.
A generation of students and their families are beginning to question whether degrees in software engineering, computer engineering, or computer science are still worth pursuing.
The reasoning is simple: if artificial intelligence (AI) can already write code, design systems, and automate technical work, then why spend four or five years studying how to do something a machine can now do in seconds?
We are already seeing the consequences of that perception. Over the past two years, applications and enrolment interest in our computer science, software engineering, and computer engineering programs have dropped by roughly 30 per cent. That is not a small fluctuation. It reflects a growing belief that these fields will become obsolete as the use of AI becomes more prominent.
It is an understandable concern. But it rests on a misunderstanding of both what AI actually does and what computer scientists, engineers, and technical professionals are trained to do.
Much of the public conversation assumes that software engineers and computer scientists simply write code. If AI now writes the code, the need for software engineers or computer scientists disappears.
But that was never the real job.
Consider the calculator. When calculators became widely available, they eliminated the need to perform arithmetic by hand. Yet, mathematicians and engineers did not become obsolete. Instead, the value of human expertise shifted to something more important: understanding the problem, framing it correctly, and determining whether the answer actually made sense.
The same shift is now underway in the development of software systems and engineering.
Recent industry analysis, including Thoughtworks’ “Future of Software Engineering” retreat findings, suggests that AI is most effective at accelerating the production of code. But the most critical work is moving elsewhere. Engineering and technical rigour does not disappear when AI writes the code. It moves upstream.
Practitioners increasingly report that oversight is shifting away from inspecting code and toward evaluating the specifications, constraints, and tests that guide AI systems. If an AI agent generates software from a specification, that specification becomes the most important artifact in the system. Poorly defined requirements can produce errors at enormous scale. Testing is also becoming central; some organizations now treat test suites as the primary engineering artifact, with AI-generating code that must satisfy those tests.
This is what practitioners call the “middle loop” of engineering work: the human layer between writing code and delivering products. It includes shaping problems, supervising AI-generated output, managing risk, ensuring security, and deciding what should actually be deployed.
The research also challenges another common narrative: that AI will eliminate entry-level technical jobs. In some cases, junior engineers may adapt more readily to AI-enabled workflows because they are less tied to existing habits and assumptions. Rather than hollowing out the profession, AI may reshape it in ways that place new value on adaptability and foundational understanding.
Degrees in software engineering, computer engineering, and computer science do far more than teach students how to write code. They teach systems thinking, mathematical reasoning, design under constraints, risk analysis, and professional responsibility. Students learn how complex systems fail, how to test and validate ideas, and how to make decisions under uncertainty. At the University of Waterloo, our co-operative education model ensures students apply that knowledge in real workplaces throughout their degree—learning not just how technologies work, but how to supervise them, evaluate them, and use them responsibly.
They also learn something even more important: accountability.
As AI systems become embedded in critical infrastructure, from health care, to transportation, to finance, the consequences of technical decisions grow larger. When something goes wrong, it will not be the algorithm that explains what happened. It will be a human being.
AI does not remove responsibility. If anything, it concentrates it.
Across all engineering disciplines, the distinctly human contribution is judgment: knowing when to trust a model and when to question it, recognizing unintended consequences and balancing efficiency with safety and public trust. Those capabilities are not shortcuts. They are developed through rigorous education and experience.
So if you are wondering whether a degree in software engineering is still worth pursuing, the answer is yes—not because the field is unchanged, but because it is changing profoundly.
AI may write the code, but computer scientists and engineers will still have to answer for it.