‘Do not focus on the end product, but make the process the main priority’

Does a thesis or exam still make sense now that AI is developing at lightning speed? According to Kai Yu Ma, academic teacher at the UMCG and the University of Groningen, language models such as ChatGPT have a profound impact on education. In this first episode of our JTS Scholars series, he talks about new forms of teaching and having an open conversation about AI with students.

Text: Jelle Posthuma

The breakthrough of ChatGPT initially caused fear among many fellow lecturers, Ma says. A technology that allows students to complete assignments or tests with the click of a button might be better off banned by the university. But the academic teacher thinks otherwise. ‘In my view, it is a technology that is revolutionizing the way we work – and this is already happening. It is our task as lecturers to prepare students for this.’

Language models can bridge the gap between our ‘messy minds’, which think in natural language, and the structured way computers operate, Ma explains. He decided to dive deeper into the technology. ‘I was one of the first to respond enthusiastically to the possibilities of AI in education. By watching videos and reading scientific articles, I learned more about AI models such as ChatGPT. As a result, I was increasingly asked to give workshops and seminars on the topic.’

The shortest way to the diploma

By now, the lecturer is regarded as an expert in the field of AI and educational applications. In his view, artificial intelligence is changing not only the subjects that are central to the curriculum, but also education itself. ‘Education has two central goals: acquiring knowledge and skills, and obtaining accreditation – the diploma. Ideally, the emphasis from the educational institution is on the first: the development of knowledge and skills. Students, on the other hand, often focus mainly on the second goal: obtaining the diploma, the ‘piece of paper’.’

Whenever possible, students seek the shortest route to the diploma, Ma continues. In the most extreme case, this means cheating, for example through fraud. ‘ChatGPT makes it possible to generate a paper or assignment. That was possible before too, for instance with cheat sheets or a ghostwriter, but the threshold has become much lower. With the help of AI, a student can gain accreditation without being able to demonstrate the required skills.’

The new technology creates distrust between students and lecturers, he says. For many lecturers, this is a reason to ban AI technology. But reliable AI fraud detection does not yet exist, and according to Ma there is no easy way to determine whether something has been generated.

Focus on the process

The lecturer therefore advocates a different approach to artificial intelligence in education. He points to the various steps within the learning process. In the past, students started with the basics: memorizing and reproducing information, and gradually worked towards creating something independently, such as a paper or thesis. With the emergence of tools such as ChatGPT, students can seemingly skip those first steps and jump straight to the final phase. That is why it is important to assess and design the learning process differently.

Within the university, discussions are underway on whether students should still write a thesis, the lecturer says. Perhaps writing a good prompt (an instruction for AI, ed.) is more important. ‘We should no longer focus on the end product, such as the exam or thesis, but on the process.’ After all, this process is indeed crucial, he continues. Students should possess the skills to critically reflect on the outcome. ‘Suppose ChatGPT gives you five options. As a student, you still need to be able to choose the best one. And if none of those options are sufficient, you need to come up with a sixth one yourself. Generating information is not difficult, but critically assessing it certainly is.’

Tailoring and open conversation

Moreover, the process of going through something yourself fosters perseverance. 'Staring at a blank screen, the discomfort that comes with it, and then creating something yourself is a valuable skill. It gives meaning to your work. In addition, writing helps to structure your thoughts and formulate your opinion. Sometimes it is good to allow ChatGPT, but AI can also disrupt the process, because students end up at the final step straight away. That requires proper guidance and tailored decisions.’

The question is how lecturers can assess and grade the process in an efficient way. ‘Assessing a final paper or exam is relatively straightforward. Assessing a process is much more complex. We don't yet know how this should be done in a world with AI.’ Above all, Ma argues for an open conversation between students and lecturers about AI. ‘Students use it a lot, but they are afraid to talk about it openly. It is up to lecturers to make it clear that some AI technologies hinder the learning process, and that it is not about the diploma or a cum laude grade, but about the skills you develop for the future.’

According to him, this ‘open conversation’ is still far too rare within the walls of academia. Education is always lagging behind, the lecturer says. ‘We work in cumbersome organizations, while AI technology is advancing at a tremendous pace. Rules for the use of AI are now in place within the institutions, but these were drawn up based on the capabilities of chatbots from 2023. By now, these rules have already been overtaken by new technology, such as AI agents, which can perform more complex tasks without human intervention.’

Improving the educational process

With the JTS Scholarship, Ma wants to further investigate the implementation of artificial intelligence in education. One example is the use of AI tools in teaching. ‘Medical students learn clinical reasoning during their training. This is a complex task, for which they now use role-playing with other students or actors. I am working on an AI simulation of this role-play to help acquire such skills. The great thing is that this technology is always available.’

He also wants to continue the debate on AI within the academic world with the Scholarship. ‘My main question is how we can use artificial intelligence to improve the educational process. This can be done with the help of AI tools, but we can also make a conscious decision not to use AI for certain skills.’