With the arrival of ChatGPT, the use of AI in education has grown rapidly, but for Theo Bakker, it’s nothing new. “I used my first AI model in education 15 years ago to predict student dropout,” says Bakker. He has a background in the intersection of data, educational systems, and IT strategies at the University of Amsterdam, Vrije Universiteit Amsterdam, and Deloitte. He earned his PhD at VU Amsterdam on the academic progress and success of students with autism. He currently works as a lecturer in Learning Technology & Analytics at The Hague University of Applied Sciences, and is affiliated with VU Amsterdam as assistant professor in Clinical Neuro- and Developmental Psychology.
Bakker has noticed that his area of interest – equity in education – and AI are increasingly mentioned in the same breath. “But the discussion often stops at the question: ‘Can students afford the necessary IT tools?’ Equity is about much more than that.” That’s why the Npuls programme decided it was time to create a vision on the impact of AI on equity in further education. Npuls is a collaborative initiative by Dutch secondary vocational, higher professional, and academic education institutions, along with the IT cooperative SURF. Theo Bakker is the editor-in-chief of the vision document, which he co-wrote with colleagues from these various educational institutions.
Equity
The document begins by clarifying what the authors mean by equity in education. The experts argue that a student’s educational journey should depend as little as possible on the environment in which they were raised. Professor of Mathematics Sandjai Bhulai writes: “To create a fair chance for everyone, we must break the vicious cycle of inequality passed down from parents to children. That sometimes requires treating students in unequal ways – such as giving more time and attention to a student with fewer opportunities than to one with more. This may sound contradictory, but it’s an important step towards achieving equal outcomes.”
How can the use of AI help with this? In the document, Bakker and his colleagues list both AI's potential positive and negative impacts. This shows that every potential advantage also has a possible downside. Bakker says: “This has always been the case with technology, and it’s something we have to deal with.” Take the impact of AI on student guidance, for example. On the one hand, AI can offer personalised guidance according to each student’s needs and pace. On the other hand, there’s a risk that AI caters to the average student and fails to assess individual situations properly.
Training AI
Another potential advantage or disadvantage is AI's widely discussed objectivity or subjectivity. “An algorithm is objective but reinforces patterns from the initial data,” Bakker explains. “Any existing bias gets reinforced, and we don't always see it because the outcomes often feel so recognisable. We can train AI and use specific prompts to take inclusivity into account.”
The authors also list several case studies of successful AI applications and negative consequences. One positive example is an AI tool VU Amsterdam and ROC TOP developed to support vocational students at higher risk of dropping out. The system uses study data and AI to identify students needing additional support. This information is then shared with mentors and academic counsellors, so that they can help these students more quickly and prevent them from dropping out.
Bakker is hopeful about the use of AI in education. The document offers institutions various tools for using AI in a careful and fair way – for instance, by involving people from diverse backgrounds in its development and use. Bakker concludes: “As with any new technological tools, you only begin to understand their true impact when you start using them. We need to learn more quickly from our experiences with AI, so we can respond in a timely and thoughtful way.”
