Dr. Jared Pyles recently attended the Association for Educational Communications and Technology conference and has prepared some reflections and takeaways for us. Join us for this mini-series on the hottest topics discussed at the conference!
As expected, there are a lot of research presentations on Generative AI at AECT. Shocking, right?
One presentation was on a study coming out of a public higher education institution in the US. They studied undergraduate students’ perceptions of using ChatGPT in higher education coursework and the ethical dilemmas associated with it. In this study, 365 undergraduate students filled out an online survey that contained demographics and presented six ethical scenarios of using GAI to various degrees in completing graded assignments commonly found in courses. One scenario was a student who was behind on completing a discussion post in an online course, and they didn’t have time to read the materials and write it themselves. In a way to save time, they used ChatGPT to help formulate the response, copying/pasting the discussion prompt into ChatGPT and then copying/pasting the response into the discussion forum.
The researchers were in the midst of analyzing the data from these questions, but what they did present was some preliminary analysis based on responses. The early results show that students consider having ChatGPT write a paper for them as highly unethical with a score of 6.73 on scale of 1-7 (1 = lowest, 7 = highest). Students ranked having ChatGPT explain reading materials as the lowest of the scenarios (2.9). Altogether, the trend leaned toward the more summative an assessment, the more unethical they view using ChatGPT to complete it. The researchers noted that the likelihood to use ChatGPT was still relatively low. They also noted that there seemed to be an ethical shift from writing to learning. Students saw that using ChatGPT to help with learning (such as summarizing reading assignments or suggesting sources and summarizing) was more ethical than using it to complete assignments that had a writing component, such as papers or discussions posts. It should be noted that the scenarios were limited to assignments involving writing since ChatGPT is a Large Language Model. Another note discovered from this study was that actual student usage of ChatGPT was relatively low although instructors may assume that students are rampantly using ChatGPT and must be stopped or prevented at all costs.
Something that wasn’t discussed but is something CTL members have thrown about in the office, is how what students’ value impacts their likelihood to use ChatGPT or other GAI to complete (or cheat) on assignments. Anecdotally, we have discussed how students expressed how they tend to cheat in the courses that they don’t value as highly as they value others. This ‘value’ could be based on the instructor, the content, the purpose behind the assignments, or whether the course is in their major or outside of it. We have no answers right now, but we are curious about this factor.
Overall, the conference was abuzz with these kinds of discussions. A lot of research is being done on perceptions and usage of GAI by students, instructors, and other practitioners, and we’ll continue discussing it in future posts.
- 2024 AECT Reflection: Student Usage of GAI, Part 2 - November 5, 2024
- 2024 AECT Reflection: Student Usage of GAI, Part 1 - October 31, 2024
- Generative AI, In Practice: Using GAI for Course Planning - June 13, 2024