Campus Internationalization or Campus Whitening?: AI Analysis of Image Data on Japanese University Websites

Abstract Scholars have raised concerns that White-centric norms have been widely prevalent in higher education globally, bolstering the existing power imbalances that have historically been favorable toward White. This study seeks to measure the extent of White influence in a non-Western country's higher education sector, utilizing artificial intelligence (AI) techniques and big data. Specifically, this study used a facial attribute recognition algorithm to annotate race/ethnicity groups for over 670,000 images scraped from roughly 1,000 Japanese university websites featuring campus internationalization. Our analysis found a stark overrepresentation of White images relative to their actual enrollment in Japan. The websites also display White individuals more frequently than Black and South Asian individuals. Our research supports the "campus whitening" hypothesis, which claims that non-Western institutions may value White more than other racial/ethnic groups in their pursuit of international diversity. It also demonstrates a novel approach to utilizing AI and big data to examine social issues in education research.