Juho Leinonen: Problems with self-regulation can cause students to rely too heavily on AI

What do you research and why?

My research focuses on AI (artificial intelligence) in university-level programming education. The research topic is interesting because AI and AI tools have become more common in the last couple of years, and as a result, their impact on teaching has increased.

Experts benefit from generative AI and are able to critically assess whether the code or text produced by AI is useful or not. The best novice programmers are quite skilled at using AI. However, students who have problems with self-regulation in particular may easily rely too much on AI. They may leave their assignments to the last minute, which may increase the temptation to use AI. They may copy code or text generated by AI without even reading it. They may also have a false impression of their own abilities due to overuse of AI, and it may become apparent during an exam that they do not know anything at all.

Artificial intelligence widens the gap between students. If you have poor self-regulation skills, you will use AI earlier and more often. There can be stress and challenges with motivation. Skilled students, on the other hand, use AI much later in the learning process. For example, they may just check with AI to see if they could have done the task better. I have also studied how this gap could be narrowed, for example, by teaching critical use of AI.

How does misuse of AI manifest itself in code?

It is certainly difficult to say for sure that AI has been used. But on a general level, there may be certain signs of AI use in the code. These may be, for example, code structures that have not been taught in the course. The code may also contain a lot of comments, which students do not usually do. Or the code may be overly optimized, for example, with no extra constructs. This is atypical for students, who often also produce non-optimized solutions.

How did you become an assistant professor?

In the summer of 2015, I was an undergraduate student at the University of Helsinki and got a summer job in Arto Hellas' research group, studying the application of machine learning in programming education. Hellas eventually became my dissertation supervisor, and now he is my colleague at Aalto.

I defended my dissertation in 2019 and continued as a postdoctoral researcher at the University of Helsinki. I was supposed to leave for Auckland as a postdoctoral researcher at the beginning of 2021 for a year, but New Zealand had strict coronavirus restrictions at the time. This delayed my departure by two years, and I did some postdoctoral research work in the meantime. Finally, at the beginning of 2023, I was able to make the trip. Waiting required a lot of perseverance on my part. My entire research career could have come to a halt if I had ended up working for a company while I was waiting.

What has been the highlight of your career?

My year in New Zealand was enjoyable because I got to see how the university system works abroad. The highlight of the year was when I gave a keynote speech with a couple of colleagues at the prestigious ITiCSE conference in the summer of 2023 on how AI affects programming education.  The conference happened to be in Turku, so I travelled to Finland on a work trip from New Zealand.

At that time, it had only been six months since the release of ChatGPT, and people were wondering how to approach it. It was great to be able to talk about my own research at this point of change. Many people said afterwards that the presentation was eye-opening. They understood that generative AI will have a major impact on programming education in the future. We cannot just continue doing what we have always done.

What do you think are the most important qualities for a researcher?

Perseverance, creativity, and the ability to see the big picture. Even if articles are rejected, you have to keep working on them.

What do you expect from the future?

I already have a couple of researchers from my previous projects in my group, but now, as an assistant professor, I can expand the group. I also look forward to teaching. All my previous roles have been research-based, and I have only taught occasionally, more for fun. Now I get to apply my research to teaching – earlier it has been applied only by my colleagues.

What do you think about the AI debate?

The extremes are the most vocal. One view is that AI is all hype and has no impact, and the bubble is about to burst. Another view is that soon no one will need to do any work at all. A third view is that AI will destroy the world.

I personally think that AI can be a useful tool in a programmer's toolbox, which is why it is important to teach novice programmers how to use it. In an ideal situation, students will know how to use AI effectively when they graduate and will be better programmers than they would be without AI.

This article was originally published on the Aalto University website on 25.09.2025