Generative AI at School - Insights from a study about German students' self-reported usage, the role of students' action-guiding characteristics, perceived learning success and the consideration of contextual factors

Deutsche Version: https://zenodo.org/doi/10.5281/zenodo.10210312

The present study is one of the first to investigate the use of generative AI, such as ChatGPT and others, in German schools. The study is explorative in nature and focuses on the use of generative AI from the students' perspective. The aim of the study is to investigate whether, to what extent and for which tasks students use generative AI. We also examine whether and what kind of relationships exist between the intensity of students' use of generative AI and action-guiding characteristics considered crucial in digitalized educational environments, such as social perception of intelligent technology, need for cognition, self-efficacy expectation, technostress, and technology commitment. In addition, we analyze how contextual factors such as social support from school and parents, as well as facets of parents' social status, are related to students' AI use frequency. Finally, we examine whether AI use is correlated with students' self-perceived AI-related learning success and how it can be grouped into higher-order concepts. 

We conducted a quantitative analysis on a dataset (N=226) collected through an online survey between March and July 2023 with students aged 15 to 19. The results show that there are differences in generative AI usage frequency for different types of tasks. It is expected to become increasingly used by students for doing homework, writing texts, and supporting creative processes such as brainstorming and research. Students' self-efficacy expectation for being able to do something useful with this technology seems to play an important role in the context of generative AI use. At the same time, we see that students' perceptions of technology-related stress can be important when using generative AI. The results also show that social support from educational institutions and parents plays an important role in the use of generative AI. In contrast, the levels of parents' education as well as academization are negatively correlated to generative AI use, particularly in the context of social media use. In addition, students' social perceptions of AI tools, especially regarding the perception of generative AI as partially human-like (anthropomorphic), seem to be relevant when using generative AI. Interestingly, a higher frequency of generative AI usage is associated with a lower level of cognitive engagement as well as belief in technological competence among students. However, higher levels of students' self-perceived learning success are associated with a higher intensity of generative AI usage. Finally, we grouped the here developed 31 types of generative AI use into four higher-order concepts of generative AI use, which we named "performing standard tasks," "exploring new opportunities," "improving one's own work results," and "inspiring creative thinking." 

With this study on the use of generative AI from the students' perspective, we aim to contribute to a better understanding of how generative AI can change young people's learning processes today and in the future. In the final discussion section, we argue that the results of our analysis can contribute to the development of sustainable approaches on ways to transform the educational system so that it empowers young people in our technologically permeated, knowledge-intensive society to become creative, reflective, and mindful citizens of the future society.