This study investigates the amplification of gender bias by online images. Analyzing millions of images and billions of words from Google, Wikipedia, and IMDb, researchers found that gender bias is significantly more prevalent in images than in text. A nationally representative experiment further demonstrated that searching for images of occupations, rather than text, amplifies gender bias in participants' beliefs. The study highlights the crucial need to address the societal impact of this shift towards visual communication to create a fair and inclusive internet.

Douglas Guilbeault, Solène Delecourt, Tasker Hull, Bhargav Srinivasa Desikan, Mark Chu, Ethan Nadler