Does AI produce biased results? A simple test.

I asked an image generator to generate photos of a CEO. I got this:

I asked for photos of a doctor, and it gave me this:

Senator?

Notice anything about these photos? AI gets its information from existing information, and it’s a pattern recognizer. Most stock shots are of white people, and most stock shots of people in leadership positions are of white men. I get that.

But the more we use AI the more biases get perpetuated. What happens when kids look up images? And these biases are overt. What happens when AI tools take an even larger role in deciding who gets a job or a loan or what neighborhoods should get priority in a crisis?

I’m by no means qualified to speak about the technical aspects of AI, but I do know a bit about human behavior and perceptions. If, moving forward, we rely on a technology that depends so much on existing patterns, that is the opposite of progress.

Rudy Fernandez is the Founder and Creative Director of Creative Outhouse.