Waiting, Banning, and Embracing: An Empirical Analysis of Adapting Policies for Generative AI in Higher Education

Ping Xiao,Yuanyuan Chen,Weining Bao
2023-05-25
Abstract:Generative AI tools such as ChatGPT have recently gained significant attention in higher education. This study aims to understand how universities establish policies regarding the use of AI tools and explore the factors that influence their decisions. Our study examines ChatGPT policies implemented at universities around the world, including their existence, content, and issuance dates. Specifically, we analyzed the top 500 universities according to the 2022 QS World University Rankings. Our findings indicate that there is significant variation in university policies. Less than one-third of the universities included in the study had implemented ChatGPT policies. Of the universities with ChatGPT policies, approximately 67 percent embraced ChatGPT in teaching and learning, more than twice the number of universities that banned it. The majority of the universities that ban the use of ChatGPT in assessments allow individual instructors to deviate from this restrictive policy. Our empirical analysis identifies several factors that are significantly and positively correlated with a university's likelihood of having a ChatGPT policy, including the university's academic reputation score, being in an English-speaking country, and the general public attitudes toward ChatGPT. In addition, we found that a university's likelihood of having a ban policy is positively associated with faculty student ratio, citations, and the English-speaking country dummy, while negatively associated with the number of peer universities within the same country that have banned ChatGPT. We discuss the challenges faced by universities based our empirical findings.
Computers and Society,Artificial Intelligence,Machine Learning
What problem does this paper attempt to address?
This paper aims to explore how universities formulate policies regarding the use of generative AI tools (such as ChatGPT) and analyze the factors influencing these decisions. Specifically, the study focuses on the following two questions: 1. **The current status and trends of university ChatGPT policies**: By analyzing the ChatGPT-related policies of the top 500 universities worldwide, the study examines the existence, content, and publication dates of these policies. The research finds that less than one-third of universities have established ChatGPT policies, with approximately 67.4% of these universities choosing to embrace ChatGPT for teaching and learning, while only about 32.6% prohibit the use of ChatGPT in assessments. 2. **Factors influencing universities to formulate different ChatGPT policies**: The study investigates which factors are related to whether universities formulate ChatGPT policies and their specific content (embracing or prohibiting ChatGPT). The analysis indicates that factors such as the university's academic reputation score, being located in English-speaking countries, and public attitudes towards ChatGPT significantly positively influence the likelihood of universities formulating ChatGPT policies. Additionally, the teacher-student ratio, citation counts, and other factors are also related to whether universities prohibit the use of ChatGPT. Through this research, the paper provides valuable academic and practical insights to help policymakers understand how higher education institutions are responding to emerging AI technologies and make informed intervention decisions when necessary. At the same time, the research results also help institutions understand the current application of ChatGPT in the field of higher education and how peer universities are formulating relevant policies to guide faculty and students in using these new technological tools.