2023-04-26 17:23:00
Floris and Sindala are in trouble. They are locked in a dungeon and must escape. A class full of high school students looks intently at the black and white images. While Rutger Hauer and Jos Bergman cleverly break out, the students answer questions about what they see. The questions were partly devised by the writing program ChatGPT, which uses artificial intelligence.
ChatGPT is a smart chatbot developed by the company OpenAI. A user can ask a question, after which the bot writes an answer within a few seconds. The program can do that very well. Harmen de Jong is a history teacher at the Topsport Talent School (TTS). “If I had come up with all the questions myself, it probably would have taken me an hour. Now I was done in twenty minutes. I only had to come up with a few questions myself and adjust ChatGPT’s suggestions.”
The program can relieve teachers of work. It is free and accessible to everyone, including students who have to write a report. This poses a problem for education. ChatGPT does not literally copy text from its database, but writes a unique text by varying on the basis of existing texts. It is therefore hardly distinguishable from self-written text. As a result, it is almost impossible for a teacher to check whether a student has worked on the assignment himself.
“Not 100 percent reliable”
Students know how to find ChatGPT, but are critical of its use. Boaz Lockhorn and Florian Zwaving are in the 3rd class HAVO/Atheneum of the TTS. They ask a question in ChatGPT and ponder the answer. Florian: “I would adjust it a bit. You can get information from it, but then you still have to do something with it.” Boaz adds: “It is not 100 percent reliable.” The boys wouldn’t just turn in a report written by the program. Florian: “If you take it literally, you won’t learn anything from it.”
It seems a foregone conclusion that software like ChatGPT will change education. Bart Verheij, professor of artificial intelligence at the University of Groningen, also thinks so. “Certain things, such as writing essays, can no longer be done in the old way.” It is not new that new technology forces education to adapt. “When I went to school you didn’t have the graphing calculator yet, there was attention for the art of drawing graphs. That has now completely changed and has made new forms of education possible,” says Verheij.
Stay critical
According to Verheij, ChatGPT will not take all the work off the students’ hands anyway. “Critical reflection is an important subject taught to students. Even with the rise of artificial intelligence, this remains very important, the user must check whether what is written is really correct. ChatGPT itself has no idea about that.”
Jiulie Buigel (left) and Isabel Schollena (right)
So you have to look at the answer with a critical eye. Jiulie Buigel and Isabel Schollena, also from 3 HAVO/Atheneum of the TTS, also notice this when they ask ChatGPT a question about the end of the Cold War. An answer rolls out, the program states that the war ended in 1991 with the collapse of the Soviet Union. “But that’s not quite right,” says Jiulie, “We have learned that you can also see the fall of the Berlin Wall in 1989 as the end.”
This is a major flaw in programs like ChatGPT, so great it even has a name. Verheij: “This is called the ‘hallucination problem’. The program can say something very convincingly, it also sounds credible, but it is actually not correct at all.”
Student laziness
When history teacher Harmen de Jong hears that his students correct the program about the end date of the Cold War, he starts to applaud. “I think it’s great that my students can look at it critically, and say: 1989 is also a good answer.” Still, he knows that students do use the software to write reports; he has already encountered this once. “That is actually laziness on the part of the students. They then took it so literally that I can see that it was written by AI.”
When a student handles it a little more carefully, the text is indistinguishable from the real thing. De Jong: “I have heard that scanners have been developed and that they are already being used at universities of applied sciences. That just doesn’t seem to be watertight.” Professor Verheij is also critical of this solution. “I do not advise anyone to go down that path. You end up in a kind of arms race. It is much more interesting to work with the new possibilities together.”
Greater role in society
The scientist predicts that artificial intelligence will play an increasingly important role in society. He talks about ‘democratization of technology’. “Techies already use artificial intelligence as a ‘regular’ tool. I expect it to become a tool for everyone.”
He does acknowledge that there are risks associated with the further integration of artificial intelligence in society. “Not everything can be prevented, there will be disasters and disasters with artificial intelligence.” Still, he remains optimistic about the future. According to him, the central question should be: “How can you ensure that these developments are good for society? We must continue to educate artificial intelligence ourselves.”
#chatbot #replace #teacher #classroom