AI bias is when AI systems make unfair choices, just like when a toy always picks one friend to play with and ignores others.
Imagine you have a robot that helps sort toys into boxes, red ones go in Box A, blue ones in Box B. But if the robot was only ever shown red toys by one person and blue toys by another, it might start thinking that red toys are better or that blue toys belong to someone else, even though both colors are just as fun.
How AI bias happens
Sometimes, people who build AI don’t think about all the different kinds of kids who will use it. If they only test their robot with red toys, it might not know how to handle blue ones, like if a kid comes in with a purple toy and the robot says, “I don’t understand this,” or even picks the wrong box.
Why it matters
AI bias can make some kids feel left out, just like when a game only favors one player. If AI is used for things like picking who gets to play first, or even choosing which kid gets extra candy, fairness becomes really important.
Examples
- A hiring tool favors men over women because it was trained on old data that showed men were more often hired.
- An AI-powered loan system gives higher interest rates to people from certain neighborhoods based on past patterns.
- A facial recognition app struggles to identify darker skin tones because the training data had mostly lighter skin tones.
Ask a question
See also
- How Do Computers Understand You?
- How Do Computers Understand Speech?
- How do large language models learn to talk like humans?
- How Do Smartphones Know You're Looking at Them?
- How do large language models like ChatGPT actually learn?