AI challenges students’ learning and trust

When new students start at university, many have already used generative AI for schoolwork. They arrive with varying perceptions of what the technology can be used for and how it affects their learning.

This is shown by the results of a study being carried out by researchers from the Centre for Upper Secondary and Vocational Education Research at SDU as part of an evaluation of the Ministry of Children and Education’s four AI initiatives in upper secondary education.

According to Professor Ane Qvortrup, the findings are relevant to universities because they shed light on the experiences and study habits that new students bring with them.

1. What do new students bring with them from upper secondary school?

– Among other things, they bring a complex sense of mistrust, which is not just about whether anyone is cheating. In our survey, 60 per cent say that many students use AI to cheat on their schoolwork. Some feel that pupils who use AI get better grades than those who do not. Others find themselves suspected of having used AI, even though they have not.

This can both create mistrust amongst pupils and strain the relationship between pupil and teacher. According to Ane Qvortrup, this relationship is of great importance to learning, because pupils need to feel confident enough to acknowledge their mistakes and engage in a dialogue about their academic development.

The survey also indicates that pupils are not necessarily aware of how AI affects their learning. 37 per cent say they find it difficult to do without AI when they have to tackle a written assignment. At the same time, only 17 per cent believe that AI tools hinder their learning.

– This raises the question of how aware pupils are of what learning requires, and what it means to work with and without AI. This could be of great significance when they start university and must read and produce texts independently.

2. When does AI support learning?

– We see positive results when pupils are supported in using AI to explore concepts, develop ideas and work on problem formulations. Here, the technology can act as an academic sparring partner and strengthen their understanding and skills.

The situation is different when AI takes over parts of the academic work. Students report that this happens particularly with homework and smaller assignments, where they feel pressed for time or are less mindful of its use than with larger written assignments.

– This is significant for universities because students there are given a great deal of freedom to work independently. If AI is used to write or summarise texts without academic reflection, the technology can take over some of the work from which the student is supposed to learn.

3. What happens when teaching fails to keep pace?

Many pupils consider themselves competent users of AI, whilst teachers do not possess the same skills.

– When AI is not properly integrated into teaching, pupils’ own experiences take the lead. It can be difficult to refuse an AI tool’s offer to draft the next paragraph if the pupil has not already taken a position on what the technology should be used for.

Upper secondary schools have drawn up rules for AI, but according to Ane Qvortrup, general rules can be difficult to translate into specific teaching practice.

– If teachers do not discuss AI with pupils, the pupils will not necessarily become aware of how the technology affects their work and learning.

4. Does AI risk widening the gap?

– It is particularly those with the weakest academic prerequisites who risk being put at a disadvantage. Our study suggests that pupils with weaker academic prerequisites and socio-economic backgrounds use AI the most and often in a less reflective manner.

Students with a solid academic understanding are better equipped to evaluate an AI-generated answer and spot errors and omissions.

– If AI is used as a learning partner, the technology can make a positive contribution to academic development. However, if it primarily becomes a means of getting an assignment written, it may exacerbate existing inequalities.

5. What should SDU do?

– First and foremost, we must create spaces where students and lecturers can speak openly about the use of artificial intelligence. When it is safe to talk about how one uses AI, it becomes possible to have a constructive dialogue about trust, opportunities and problems.

Guidelines are necessary, but they must be as specific as possible. Ane Qvortrup also recommends subject specific teaching on AI at faculty, programme or subject level.

– Students must learn to use AI in a way that supports their own thinking, writing and judgement. If the teaching is too general, they will not necessarily receive the help they need to link the technology to their academic work.

At the same time, there is a need for professional development for lecturers.

– It can be confusing for students if they encounter widely differing requirements and approaches. As AI can influence learning both positively and negatively, there is a risk that inappropriate use will prevail if the university does not take an active role in its development.