Facial recognition systems misidentified Black faces at 10 to 100 times the rate of white faces. Same systems. Same study. Sold to law enforcement. Sold to the courts. Sold to the same system Tupac was already talking about in 1994.

Healthcare algorithm deprioritized Black patients for care. Not because someone coded discrimination. Because the model used cost as a proxy for medical need — and Black patients had historically been given less care and less spend. So the algorithm learned they needed less. The bias was baked in before the first line of code.

Hiring algorithm trained on ten years of resumes — from a decade when the company hired mostly men — learned to penalize resumes that included the word "women's" (as in women's chess club, women's college). Amazon scrapped it. The model was doing exactly what it was trained to do.