⚖️ AI Bias
5th Grade 🤖 Artificial Intelligence
Bias in AI is when a system treats different groups of people or things unfairly or unequally — usually because the data it learned from was one-sided.
A system learns from the examples it’s given. If most pictures of “doctor” in the data show men, the system might “learn” that a doctor is always a man. If a face recognition system was trained mostly on faces from one group, it may make more mistakes with people from other groups. The system doesn’t “want” to be unfair — it just reflects what was in the data.
Bias can cause real harm — for example, if a system helps decide who gets a loan, a job or medical care. That’s why responsible developers test their systems on different groups, add more varied data and have people review important decisions.
A common mistake is thinking a computer is always fair because “it’s just a computer.” A computer can repeat the prejudices that were in its data, and even make them bigger. You can spot bias too: when you notice an AI tool always gives the same kind of answer or picture, it’s worth asking — why?
For example
- An image generator that shows only one kind of person when you ask for “a scientist.”
- A face recognition system that makes more mistakes with people who barely appeared in its training data.
- A recommendation system that suggests only football games to boys and only dance games to girls.
- A speech recognition system that understands an accent less well because it rarely heard that accent in its recordings.
📌 Remember: Bias = one-sided data → unfair decisions. Variety and testing help.
Quick quiz
What is a common reason for bias in AI?
Is a computer always fair?
How can bias be reduced?