How a finance professor’s engagement with ChatGPT created more authentic assessments–and more trust
After hearing about ChatGPT in the news, Andrew Hession-Kunz, Finance Faculty at Carroll School of Management, reached out to CDIL to try to mitigate cheating for his fully-remote course.
When he put his final exam questions into ChatGPT, he was shocked to discover that it answered 24/25 of them correctly. All mathematical questions were answered both correctly and precisely. Written responses were more generic and less specific than required to demonstrate full understanding of the material for the course.
In consultation with CDIL staff members and then later on his own, Prof. Hession-Kunz, experimented with his pre-existing final assessment and ChatGPT. He explored how student use of GenAI might undercut the learning goals for his course, as well as how the technology could be useful.
Based on this experience, he revised his final assessment for his course to be more authentic and representative of what students might encounter post-graduation. His new version required students to identify a particular company type they are interested in and then engage with ChatGPT, asking via the chatbot questions about content related to the course. Students then analyzed the output that ChatGPT generated and critiqued it, referencing specific course content.
Prof. Hession-Kunz felt very positive about the changes he made to his assessment and reflected that the process caused a shift in his relationship with his students, requiring him to “trust [them] tremendously.” When reflecting on the impact this change had on his course, Prof. Hession-Kunz shared, “I certainly felt that it was more authentic to acknowledge that ChatGPT exists, rather than just ignore it… In my experience, and across history, prohibitions by fiat are largely unsuccessful. ChatGPT is so magnetic, so alluring, that to just ‘ban its use’ seems destined to fail.”
For more resources related to teaching and AI, our section “Engaging with Generative AI.”