University of Washington researchers found significant racial, gender and intersectional bias in how three state-of-the-art large language models ranked resumes. The models favored white-associated...