noun
AI model bias refers to systematic errors in artificial intelligence systems that lead to unfair or prejudiced outcomes. These biases often stem from imbalanced training data, flawed algorithms, or unintended design choices that favor certain groups over others.
For example, a facial recognition system trained mostly on light-skinned faces may perform poorly on darker-skinned individuals, leading to higher error rates. Similarly, a hiring algorithm trained on past hiring data may discriminate against certain demographics if historical hiring practices were biased.
For developers and users, AI model bias can lead to legal risks, reputational damage, and ethical concerns. Companies must audit their models, diversify training data, and implement fairness checks to mitigate these biases. A 2020 study found that some commercial AI hiring tools were 2.5 times more likely to reject qualified female candidates than male ones.
A language model trained mostly on English texts may struggle to generate accurate translations in less common languages.
In other languages
See also: Ethical AI