Teacher Students Interpret Poems with AI. On Potentials, Risks, & Disruptions in L1 Teacher Education
Submitted by:
Marco Magirius
Abstract:
ChatGPT and other chatbots will soon become ubiquitous companions unsettling all professional areas of life. This also applies to the preparation of literature classes. In our research report, we present results from two studies that inquire the research question: What risks and potentials can be identified when teacher students use ChatGPT for interpreting poems?
In the first, exploratory study (Anonymized, in prep.) we asked teacher students from four seminars (n=21) at the University of Education Heidelberg to provide interpretations of literary texts commonly used in teaching and to assess their potential for literary learning. Then, they were given the same tasks again, but this time they were allowed to consult chatbots like ChatGPT for assistance. Furthermore, we discussed with the teacher students the opportunities and risks associated with the use of artificial intelligence. These discussions were recorded.
A qualitative content analysis (Kuckartz, 2018) revealed that chatbots can help students to generate various interpretative approaches and gain ideas for learning objectives. However, students often uncritically adopt interpretations from the AI. In some cases, they even fall behind their initial understanding of the texts (Anonymized, in prep.). If the students had difficulty finding interpretation ideas without AI, they also faced challenges in effectively utilizing the AI (ibid.).
Currently we are planning a follow-up study which will be completed in March 2024. We want to test a central hypothesis which we found with the first study: The ability of unleashing the potential of chatbots for the mentioned tasks depends on the students’ pre-knowledge regarding interpreting literary texts. Thus, in a follow-up study with students from Heidelberg and Berlin, we want to control two variables—the difficulty of the texts and the students’ pre-knowledge—to compare the student’s (rated) results via Mann-Whitney U tests. This way, we combine qualitative and quantitative methods (Kuckartz, 2014) in order to test our central hypothesis. Subsequently, we will outline consequences for teacher education.
References
Anonymized (in prep.) […].
Kuckartz, U. (2014). Mixed Methods. Wiesbaden: Springer VS.
Kuckartz, U. (2018). Qualitative Inhaltsanalyse [=Qualitative Content Analysis]. Weinheim: Juventa.