ChatGPT: Early reflections on how AI can be incorporated into a course

Rolf Viervant (ESPhil) discusses AI in education based on his experience
DALL·E generated image in which the Erasmus University is threatened by a robot army
DALL·E generated image of Rolf Viervant in a Carel Willink painting with robots

On the 9th of March, Rolf Viervant, a PhD candidate at ESPhil, hosted a lunch session at the Education Lab on the topic of AI in education. During the event, Rolf discussed how he implemented ChatGPT in one of the assignments of a course he teaches. Continue reading to learn more about how he decided to incorporate AI, the benefits and limitations of AI in an educational setting, and his suggestions for using AI based on his course.

The decision to use AI

Rolf recalls how he discovered ChatGPT during the Christmas holidays of 2022. It didn’t take long for him to start testing the limits of the AI to see what it is capable of, as many of us did when we first discovered it. He started playing around with different prompts, asking the AI to produce a summary of a philosophical text that his students were also required to read and summarise as part of an assignment. "It wasn't an amazing summary, but it wasn't terrible either - it was very similar to what students usually write when they first read the text," Rolf explained. This gave him the idea that if the AI could write a summary like students do when they first start studying a text, then why not use it to help students improve their summaries and general understanding of philosophical texts? With this in mind, Rolf redesigned a three-part assignment in which his students had to use ChatGPT.

Keeping educational value when using AI

The main concern that Rolf had going into this redesign was to keep the educational value of the course at a high level. The assignment was designed to teach students valuable summarising skills, so he kept this in mind when incorporating ChatGPT. The first part of the assignment focused on students finding the best prompt for ChatGPT to produce the highest quality summary. The second part focused on critically evaluating a summary provided by ChatGPT, and the third on synthesising a substantiated text. Throughout the three parts of the assignment, Rolf asked his students for feedback to learn more about their experience of using the AI.

DALL-E generated oil painting of a student who is half human half cyborg

Assignment and student response

For the most part, students were enthusiastic and interested in using ChatGPT for the assignment. “There were a couple of students who were hesitant about the programme”, Rolf mentioned, “but they were a small minority who were concerned that this would replace the standard working process”.

In the first part of the assignment, students were shocked at how well the AI was able to produce a complete summary of the text within the strict word count of the assignment. But during the second and third part, as the students began to study the text more and understand it better, there was a shock in the opposite direction- students were shocked at how many mistakes the AI would make. “The biggest drawback is the implicit assumptions that the AI incorporates”, Rolf explains. “If a student doesn’t know the text that well, they can’t spot these biases – but the more they know the text, the easier it is for them to see how an AI like ChatGPT easily incorporates these biases into its output.”

Takeaways and next steps

Overall, Rolf believes that AI and programs like ChatGPT are here to stay. He does see some significant drawbacks in them, such as the implicit assumptions and wrong information it provides at times. However, he believes that we cannot simply ignore and look away from such developments. We need to actively look for the best ways to use them in an academically sound manner. When asked, he says he would like to keep it as part of his course next year. However, he is aware that AI is developing and changing at an extremely fast pace, so he is unsure of how it will look to incorporate it into a future course.

Overall, he believes that there is a lot of potential for AI in education, across faculties and disciplines, based on the broad (although sometimes flawed) knowledge it provides, as well as the academic skills that are developed in learning how to use and work with this type of programme. He’s not sure that it’s great for a deep understanding of a subject matter (which is often needed for philosophy courses) but it’s great for a general understanding to help students get started.

Compare @count study programme

  • @title

    • Duration: @duration
Compare study programmes