Definition
Bias in AI occurs when systems produce skewed or unfair outputs, typically because of imbalances in training data or problematic patterns the model has learned to replicate. Training data bias means skewed input data produces skewed outputs. Output bias means the model amplifies patterns in ways that disadvantage certain groups.
Example
A hiring model trained primarily on resumes from one demographic unfairly favors that group in candidate recommendations, even when qualifications are equivalent.
Why it matters
Bias in AI agents can cause legal liability, reputational damage, and unfair outcomes. Testing for bias and implementing mitigation strategies is essential for responsible AI deployment.