How does artificial intelligence affect students’ learning and academic performance? This is what researchers in the Higher Education Learning with AI (HELAI) project at NTNU are investigating. AI LEARN researcher Lise Vikan Sandvik is among the researchers involved in the project.
The researchers examined the use of AI among 160 first-year students across three programming courses at NTNU in spring 2026. The results show a correlation between how students use AI and the grades they achieve in their exams. Students who most frequently turned to AI when facing challenges had an average grade of 3.2, roughly equivalent to a D. Students who used AI but were less dependent on the technology achieved an average of 4.2, approximately a C. Students who reported not using generative AI achieved the best results, with an average of 5.1, approximately a B.
The differences were statistically significant, even after the researchers accounted for factors such as previous grades, programming skills and educational background. However, this does not necessarily mean that AI use itself leads to poorer learning outcomes. It may also be that students who experience academic difficulties are more likely to seek support from AI tools. Nevertheless, the findings highlight an important educational challenge: students who use AI to obtain quick answers rather than working through problems themselves risk missing out on valuable learning opportunities.
The results highlight the need for further research into which forms of AI support genuinely enhance learning, and how higher education institutions can facilitate the effective and appropriate use of the technology.
Read more:

