U.S. college students’ acceptability and educational benefits of ChatGPT from a digital divide perspective

ChatGPT has diffused widely and rapidly, with diverse positive and negative implications. In educational settings it is important to understand students’ perceptions of the acceptability of ChatGPT for various learning activities and to examine whether prior digital divide concerns pertain to this digital innovation, in order to provide guidance for users, inform policymakers and other stakeholders, and extend digital divide research. The purpose is to investigate the associations of socioeconomic status (SES, both family and student), gender, and race/ethnicity with students’ perceived acceptability of common ChatGPT activities as well as their opinion about how beneficial ChatGPT is for college education, and additionally, whether such relationships differ based on adoption experience. We analyzed survey data quantitatively in two phases (N=360 and 1,267), applying measurement reliability and validity, correlations, and structural equation modeling. The results indicate that students with higher family SES and lower individual SES, and females, tend to view the acceptability of ChatGPT uses more positively, although racial/ethnic minorities are more critical of displacement activities. ChatGPT adopters perceive two dimensions of ChatGPT activities (academic support and academic displacement) as more acceptable than do non-adopters, and they also perceive uses for academic support more positively than for displacement. Moreover, adoption is a significant moderator of some of these associations. At this early stage of ChatGPT diffusion, these digital divide influences on acceptability and general opinion are weak and variable. The discussion further considers theoretical and practical implications for digital education in the AI era. • Studies U.S. college students’ perceptions of acceptability of ChatGPT for education • Applies digital divide perspective, focusing on SES, gender, race/ethnicity, adoption • Analyzes data from two studies (pilot N=360; full N=1267) with SEM • Finds two acceptability dimensions: academic support and displacement • Adoption slightly moderates relationships between digital divide & acceptability