Identification
- Label (rdfs)
- Harmful Bias
- Preferred Label
- Negative Bias
- Alternative Labels
- Discriminatory Bias
- Identifier
- N/A
Definition and Examples
- Definition
- Harmful bias in an AI system refers to biases that create negative impacts, such as unfair or discriminatory outcomes. These biases can emerge from various factors, including human or system decisions across the AI lifecycle, pre-existing cultural and social biases in training data, technical limitations like non-representative data, or the use of AI systems in unanticipated contexts