History and Health

We are beginning to see artificial intelligence and machine learning sweep across the medical field, changing the way healthcare decisions are made. Though research continues to show that training machine learning models on large data sets can identify at-risk patients comparable to or even better than human doctors, research also demonstrates that artificial intelligence and machine learning can and do reproduce bias and discrimination. For example, medical algorithms place Black kidney patients lower on the kidney transplant referral threshold than white patients, and some of the most widely used health algorithms identify Black patients as less ill than comparable white patients, and therefore do not refer them to programs for complex and significant health needs.