Han van der Maas | Because of AI, traditional school subjects have run out of steam
With the advent of AI, the current education system could well be shaken up, says columnist Han van der Maas. ‘My daughter in secondary school is learning exactly the same things I learned in the early 1980s when the Internet did not exist and AI was in its infancy.’
What is happening to us and can we ban it? That's pretty much the reaction of the education world to the advent of ChatGPT. I too succumbed to this. Our exam “Programming in R” is a piece of cake if students install ChatGPT as an add-in in Rstudio. So we use UvA computers for the exam on which you are not allowed to install anything. This defensive reaction is not adequate, as ChatGPT does not stand alone.
We all have cell phones on which we can find facts from any field of science in seconds. With WolframAlpha, we can easily solve math exam items. With Google Maps, knowledge of topography is superfluous. We can write poems with GPT3, compose music with AIVA, play chess with Stockfish and translate with Deepl. The list is long and growing by the day. The existential question for us is what these developments mean for education.
Especially in primary and secondary education, this question is routinely ignored. Technology has influenced how we learn but hardly what we learn. We use computers to optimize educational processes and software to memorize things faster. But when it comes to the what question, we pretend that computers and the Internet do not exist.
Elementary students learn topography the way I learned it half a century ago: memorize lots of provinces and countries once, take tests, and forget about it. The question we need to ask is what you need to know about topography in order to understand the world and use Google Maps. This is an empirical question, we need to investigate this scientifically.
Arithmetic
We also need to answer this question for math and arithmetic. Everyone who actually uses math in research or work uses sophisticated software. What do you need to know about algebra to apply this software to integral calculus? Whether you have to spend a year in high school solving nasty integrals as if this software did not exist is highly questionable.
For arithmetic, this fundamental discussion has been going on much longer. Why do elementary school children still do hundreds of sums like 234x534? The most serious answer I got was that this is how children learn to deal with algorithms. But then why don't they just learn programming in, say, Scratch? Education is beset with indirect learning arguments anyway. Students learn Latin and Greek because it makes it easier for them to learn French or Italian. But there is little evidence for this kind of transfer of skills. For example, from playing chess you learn only chess and nothing else. If you want to teach young people French then you have to teach them French. And whether you still need to teach them French is highly questionable.
Language education must also be thoroughly rethought. It won't be long before we can have a serious conversation with any other earth inhabitants with minimal delay, via our phones where each one talks in his own language. You can even do that with your own voice. There are many complaints about the youth not reading anymore, but writing and reading will presumably soon be replaced by speaking and listening. Perhaps illiteracy will be solved not by better teaching of reading and writing but by its abolition.
Teaching subjects such as biology and geography also need structural revision. The curriculum is now determined by interest groups and has not been fundamentally revised in half a century. My daughter in secondary school is learning exactly the same things I learned in the early 1980s when the Internet did not exist and AI was in its infancy. What basic knowledge about these subjects is necessary to gain specific knowledge when it is needed? Surely the most important are skills to separate sense from nonsense, check sources and arrive at a clear understanding.
The nice thing about teaching at the university is that education is adapted much faster to the demands of the time. Courses and even entire educational programs disappear when there is no longer a need for them. We think more about what our students will really need when they become researchers or start working in practice.
We now need to re-examine our programming education. Coding becomes less relevant, checking code and writing good code prompts becomes more important.
Writing education at universities must also be reshaped. It is impossible and undesirable to ban AI writing tools. By 2023, every student can effortlessly generate an essay in good English or Dutch. So that learning goal can be dropped. What do we actually want students to learn?
Han van der Maas is professor of psychological methods theory.
