Felienne Hermans began programming as a teenager – not just for herself, but also for friends, her mother and the computer club at her high school. Finding a solution for people, dealing with feedback, improving the programs – it was these human elements that motivated Hermans during her first steps in the programming world.
“When I started studying computer science at university twenty years ago, computers were already used for grading work. As a student, you don’t think much about it, but when I became a teacher, I started considering the pros and cons more carefully,” says Hermans. “For example, an automated grading program might mark an entire answer wrong because of a misplaced full stop. A teacher, on the other hand, could follow the student’s thought process and perhaps award half marks.”
Using algorithms for grading work can be disadvantageous for both teachers and students, Hermans argues. Teachers lose insight into the student’s learning process, while students may lose the idea that programming is a human process – a creative process made for and by people. Precisely what motivated Hermans as a teenager.
AI in education
When Hermans was a student, computer-based grading was mostly limited to fields like computer science. Today, she sees digital tools being used across many disciplines – and not just for grading. With the rise of artificial intelligence, algorithms and automation in education have become hot topics. Now a professor at Vrije Universiteit Amsterdam, Hermans hears the question more and more often: how should we approach AI in education?
She notes the irony. “For years, I’ve been saying that everyone should learn to program and understand how algorithms work. But people kept saying it wasn’t necessary. But now that we have ChatGPT, suddenly everyone is saying that students need to understand algorithms so they can use AI!” But that, Hermans clarifies, was not her point. “I want people to understand the technology so they can learn to create things themselves – not so they blindly use ChatGPT without being able to judge whether the output is accurate.”
Taskforce
That’s why Hermans is glad to be part of the VU-UvA taskforce on AI in education, where experts are exploring how to deal with technological developments in education. “When AI is proposed as a solution, the first question we should ask is whether we actually want it. We shouldn’t just accept AI without question; we need to carefully weigh the pros and cons first.”
Hermans emphasises the importance of the human element in education – especially the relationship between teacher and student. “When you write an essay, you’re writing it for the person who will read it – in this case, your teacher. In a tutorial group, sometimes you raise your hand to ask a question just because no one else is doing it, and you don’t want the teacher to feel bad. Even though you’re writing or asking something for the benefit of the teacher, you still learn something from the process. These human interactions help you engage with the material. We all have that one teacher who sparked our interest in a subject or motivated us to work that little bit harder.” That’s why Hermans warns against introducing AI to take on the work that teachers do. “Because that would risk losing a valuable part of the student-teacher interaction.”
Deskilling
In an upcoming article, Hermans links current developments to Harry Braverman’s theory of deskilling. Dating back to the Industrial Revolution, the theory describes how technological advancements led to the loss of certain skills. While historically a shoemaker would craft an entire shoe, with the rise of assembly lines many workers were replaced by machines. Those who were left were only responsible for a small part of the production process. As a result, much knowledge and expertise was lost.
That’s a problem we may well encounter again, says Hermans. “When we prepare a presentation or write an essay, we absorb knowledge from various sources. That knowledge serves as inspiration and can lead to new ideas at a later stage. But if we start outsourcing presentations and essays to algorithms, aren’t we deskilling ourselves?”
Students often say they want to learn how to work with AI. "But what does that actually look like?" Hermans wonders. "Take ChatGPT, for example – mistakes are always slipping in there. How do you learn to navigate that? As a teacher, the best way I can think of is to first give students as much factual knowledge as possible, so they’re protected against the misinformation that computers might feed them. If you know the facts, you can recognize and correct mistakes yourself.”
