Case Study: University of Chicago advances children’s book representation research with Google for Education

University of Chicago measures representation in 1,000+ children’s books with Google for Education

The University of Chicago, specifically researchers from its Harris School of Public Policy and the MiiE lab, faced the challenge of manually measuring representation of race, gender, and age in over 1,000 children's books, a process that was too labor- and time-intensive to be practical.

Using Google for Education's Google Cloud Vision AI, the team trained machine-learning models to automatically identify and analyze characters in text and over 200,000 images. The solution from Google accelerated their research, detecting three times more faces than other tools and achieving over 93% precision. This allowed them to cost-effectively publish their findings, which revealed significant underrepresentation of people of color and females in mainstream children's literature, generating widespread media interest and demonstrating the potential of AI for social good.

University of Chicago