Generative AI makes pedagogical expertise more important than ever

Generative AI is here to stay, and the worst thing we can do is leave students alone with technology. For that reason, the role of the educator is more relevant than ever before. 

This is one of the central conclusions that Christopher Neil Prilop draws from his research when reflecting on how generative artificial intelligence is affecting university teaching and learning. 

“No other technology has had such broad implications for academic work before. In the past, teachers could easily say: ‘I do not use technology in my teaching, so it is not relevant to me.’ With generative artificial intelligence, however, even if I do not use GAI in my teaching, I still need to understand a great deal about technology from a pedagogical perspective. Students will use it regardless of what I think about it,” he explains. 

Christopher Neil Prilop is not concerned by discussions about whether GAI will replace teachers. On the contrary, he argues that pedagogical expertise has become more important than ever. 

“My take on that is we actually need teachers more than before. But we need teachers with a very high level of pedagogical expertise. You must think carefully about how to create learning opportunities that are engaging and designed in ways that discourage students from using GAI in ways that undermine their learning,” says Christopher Neil Prilop. 

Outsourcing or cognitive partnership 

Previous research suggests that technology does not, in itself, determine students’ cognitive development. Rather, effects of tools such as ChatGPT and other large language models depend on how they are integrated into learning activities and how students use them. 

For this reason, attention should be directed towards students’ behavior when using GAI. 

“Students’ use of generative artificial intelligence can be understood as a spectrum. At one end, they may outsource the entire task to a chatbot; at the other, they may engage with it in a constructive dialogue, forming a kind of cognitive partnership. If I choose the first approach, it will be detrimental to my learning. If I choose the second, it can lead to better learning outcomes,” explains Christopher Neil Prilop. 

Teaching should take this into account and build on existing pedagogical knowledge. 

“We need to think carefully about how we design our teaching to accommodate this new reality. Unlike previous technologies, which did not fundamentally intervene in academic practice, we now need to teach students how to use this technology constructively and responsibly. That is our responsibility as educators.” 

According to Christopher Neil Prilop, the issue is less about AI literacy and more about teaching students how learning happens. 

“Students should, of course, learn the basics: what GAI can do and what it cannot do. But more importantly, they need to learn how it can support their learning. Without that knowledge, they will most likely use it for outsourcing. That is what chatbots are designed to do: provide quick answers. However, research shows that this does not promote student development. That is not what learning is about.” 

An emerging academic practice 

Generative artificial intelligence has created possibilities that did not previously exist. Students can now participate in feedback processes without interacting directly with peers or educators. In this way, technology adds an extra layer to the learning process and creates opportunities for expanded learning. 

“We are seeing the emergence of a new academic practice. Previously, when students wrote a text, most of them would write it as well as they could and submit it. Now they write the text, give it to a chatbot, and reflect on the feedback before submitting it. In that way, they expand their learning space and engage more deeply with the task,” explains Christopher Neil Prilop. 

While the learning ecology that includes AI is new, research on feedback and learning processes has a long history. The challenge now is to build on that knowledge. 

“We need to be deliberate about integrating GAI in ways that strengthen what we already know works. The AI practices we model and encourage in teaching will inevitably influence how students use the technology outside the classroom as well,” says Christopher Neil Prilop. 

Adapting chatbot behavior 

In a recent study led by  PhD students Ida Bang Hansen and Rasmus R. Hansen, students were encouraged to reflect on the use of customized chatbots. As part of the study, the research team designed two chatbots with different behaviors and asked a group of Media Studies students to interact with them. 

One group interacted with a chatbot that had been prompted to behave instructively by providing specific, action-oriented feedback. 

The second group interacted with a chatbot designed to provide metacognitive feedback. Unlike the instructive chatbot, it did not suggest direct revisions to students’ texts. Instead, it asked questions that encouraged students to reflect on how their texts could be improved. 

Afterwards, the students discussed their experiences. 

“At first, most students thought the instructive chatbot was better. Those who used the metacognitive chatbot were somewhat frustrated. But by the end, several concluded that the metacognitive chatbot might actually be more valuable because it encouraged deeper thinking rather than offering a quick fix,” says Christopher Neil Prilop. 

The approach can be transferred to teaching self-regulated learning and help students gain experience with productive feedback dialogues involving chatbots. 

“This could be a very practical way of introducing the topic in teaching. Show students two different ways of using chatbots and then let them discuss the differences afterwards,” says Christopher Neil Prilop. 

Motivation and resources matter

When students choose the easiest solution, it is often because they lack motivation. According to Christopher Neil Prilop, the best safeguard against detrimental outsourcing of tasks to GAI is therefore to focus on motivation. 

“As teachers, we need to design tasks that are meaningful above all. If I can see the value of a task and feel motivated, I want to complete it. At universities, we are in the privileged position that students have chosen to be here. They should arrive with a considerable degree of intrinsic motivation,” he explains. 

The same applies to examinations and GAI, a topic that has generated significant debate and change in recent years. 

“In general, I do not believe that students want to cheat. If students are motivated, it does not matter whether an exam could potentially be outsourced to GAI. We need to consider which forms of assessment are meaningful for the program, rather than simply eliminating all written exams or take-home exams,” says Christopher Neil Prilop. 

In addition to motivation, lack of time may also influence whether students choose the easiest solution. 

“CED recently ran a pedagogical course in which a lecturer wanted to include a peer-feedback exercise. She was concerned that students would ask a chatbot to write their peer feedback. We ended up solving the issue by giving students dedicated time to complete the task during class,” Christopher Neil Prilop explains. 

The example showed that students were more likely to complete the task without using a chatbot when time was specifically allocated to it. 

A major task: ensuring competence development 

Although recent efforts to integrate generative artificial intelligence meaningfully into university teaching have been intensive and transformative, the work is far from finished. 

“There is no doubt that it is a major task for universities and other educational institutions to ensure that teachers receive the right competence development, both now and in the years ahead. Teaching competencies are crucial in the situation we are facing,” says Christopher Neil Prilop. 

As a teacher, you can reflect on your own pedagogical skills and seek support where needed. However, the responsibility is also organizational. 

“Any integration of AI into teaching should be a pedagogical decision, not simply the addition of technology for technology’s sake. As an educator, you need to trust your pedagogical expertise. And remember that you are not alone. You have colleagues to collaborate with and leaders who are responsible for defining and advancing the overall strategy,” says Christopher Neil Prilop. 

Further knowledge

At AU Educate, you can learn more about how GAI and chatbots are used in teaching: GAI and chatbots. 

If you would like a practical introduction to generative AI, you can participate in the course Introduction to Generative AI. 

You can also book individual guidance through CED: Book a free CED consultant. 

This English text was machine-translated using ChatGPT. The author subsequently reviewed and edited the translation manually to ensure that the meaning, tone, and subject-specific terminology are rendered accurately and naturally in the target language.