Spontaneous Commentary on ChatGPT Output as an Approach to Lecture

Submitter: Laura Dumin, U of Central Oklahoma

——————————————————

The experiment:

On the first night of a graduate student seminar on “how to be a graduate student” we asked ChatGPT what to know to become and be a good graduate student. I then lectured off of the results.

I had no idea what ChatGPT was going to bring back, as this was an unplanned moment based on a student question. But the list was quite extensive and allowed me to talk for about 30 min. We looked at the responses on the list and I gave more detail and nuance to the bullet points. I also corrected ChatGPT and was clear about the corrections as I talked with my students.

Results:

I was comfortable taking this spontaneous AI literacy opportunity because it was a low-stakes activity and we could learn together about what ChatGPT said about grad school. I also knew enough about the topic to know where ChatGPT might be wrong.

The students had mostly not used any LLMs at this point and were impressed that ChatGPT quickly came up with a decent list of information in response to the question. I liked trying this activity with them because it also seemed like a good introduction to how LLMs work. It can be a tad dry and also daunting to just discuss LLMs without also showing what they can do.

Contact:

• Website: https://ldumin157.com/
• Medium: @ldumin157
• LinkedIn: @laura-dumin157