The AI Dilemma

   CHAPTER 5   

Confront and Question Bias

You should not face discrimination by algorithms, and systems should be used and designed in an equitable way.

—Blueprint for an AI Bill of Rights

Solutions architect Sean Gayle is one of the inventors of the EZ-Pass. His machine learning skills are self-taught; his degree from a historically Black college is in psychology. He says, “The problem with bias in machine learning is GIGO: garbage in, garbage out. For example, I worked with one large company that used machine learning to determine people’s ages from their faces. Invariably for the Black faces, they got the ages completely wrong—by decades. The problem started with the hundreds of thousands of images that were entered into the system as input. ...