Bias in AI refers to systematic errors or prejudices in data or algorithms that lead to unfair or inaccurate outcomes. It occurs when certain groups or perspectives are favored or disadvantaged, often reflecting existing social inequalities.
Characteristics
– Can arise from biased training data, flawed assumptions, or design choices
– May affect model predictions, recommendations, or decisions
– Often unintentional but can have significant ethical and practical consequences
– Difficult to detect without careful analysis and testing
Examples
– A facial recognition system that performs poorly on certain ethnic groups due to underrepresentation in training data
– A hiring algorithm that favors candidates from specific schools or backgrounds because of historical data biases
– Language models generating stereotypes or offensive content based on biased text sources
