“We have to integrate AI in our education”

In December, a well-attended seminar on AI and assessment was held, attracting 196 registered participants. This is just the beginning of an upcoming training campaign at Uppsala University. A task force led by Deputy Vice-Chancellor Coco Norén has conducted a horizon scanning exercise and interviewed lecturers, students and staff in the University Administration. The results show that there is great variation, as well as a lack of knowledge about AI.

“Generative AI is here to stay – I think everyone is aware of that. It means we have to review and quality-assure our assessments, and must ensure that our courses and programmes are relevant,” Bengtsson says.

Plenty of strong opinions

He acknowledges that AI is an emotive issue, on which many people have strong opinions. Having said that, it affects all staff and students at the University, so it is essential to keep up with developments.

“We have an obligation to prepare our students for their future professional roles. Many of them will be using AI in their day-to-day activities and need to develop a feel for the kind of help the technology offers but also when to avoid AI – this needs to be practised during their education.”

In the interviews, lecturers expressed great concern about the issue of assessment. How can they assure quality? How can they tell who is writing?

“Generative AI has made these issues more complex, particularly in the case of take-home exams and major written assignments. This has made it more difficult for lecturers to design and evaluate assessments in a reliable way,” Bengtsson says.

Increasing AI literacy

This is why training measures are planned, starting in the spring, to increase AI literacy among lecturers and students.

“It also follows from the Artificial Intelligence Act: if the University makes AI systems available, we also have to ensure that everyone can use them in a responsible manner.”

The idea is to offer a lecture on the basics, followed by a lecture that goes into greater depth and is based on the educational context participants are involved in.

What are the challenges?

“There’s a lack of time, many people have a lot to do, so they may not put the time into learning more about the system. So lack of time is a factor that prevents people learning more.”

AI use varies

The variation in the use of AI in different programmes and courses partly derives from differences between lecturers, in terms of their knowledge of and attitudes towards AI. Whether positive or more cautious, the attitude of the lecturer has a direct influence on how the students use the technology.

Drawing up a set of rules for students is another challenge.

“Setting up some kind of traffic light system – green, amber or red – isn’t good enough, students can get round it. I believe we need to reconsider our courses and programmes and perhaps fundamentally rebuild the way assessments are done.”

Increasing students’ knowledge is one way to get them to use generative AI in an appropriate and responsible manner. But on top of that, programmes also need to review the content and design of their courses to make them relevant and prepare the students for their future professional activities.

“You can set authentic tasks, perhaps reflecting working life or future professional activities. That makes it more relevant for students to actually carry out the assignment themselves.”

Reasonable deadlines

It is also a matter of setting the students reasonable deadlines. If they are asked to deliver a five-page assignment the day after tomorrow, that may not give them a fair chance to work on the material themselves.

“Lecturers can also say that the students are allowed to use AI but it has to be transparent, and they themselves must master the knowledge and will then be tested on the knowledge. Using AI as an additional layer can actually accelerate or increase knowledge acquisition if it’s used responsibly,” Bengtsson says.

Annica Hulth