How do AI systems develop gender bias and stereotypes?

AI systems learn from examples, just like kids learning from stories. When AI is trained using information that has gender bias, like saying boys are good at math and girls are good at art, it starts to believe those ideas too.

Imagine you're watching a robot learn by looking at pictures of people doing jobs. If most of the time, you see men working as engineers and women working as teachers, the robot might think that’s how things always are. It doesn’t know anything else yet, so it starts to believe that men are engineers and women are teachers.

Now imagine this robot is helping pick who gets a job or which toy a child should get. If the robot only sees men as engineers, it might think a girl isn’t right for the engineering job, even if she’s really good at it! That's how gender bias and stereotypes can grow inside AI systems.

It’s like learning from a book that always says boys play with trucks and girls play with dolls, you start to believe that’s all there is, until you see other possibilities.