Alignment & Safety
Algorithmic Bias
When an AI system produces unfair or skewed results, usually because its training data reflected existing bias.
Algorithmic bias is systematic unfairness in a models outputs, often inherited from patterns in the data it learned from. A hiring model trained on past decisions can absorb and repeat the prejudices in those decisions, for example. Because the bias hides inside a complex system and can look objective, detecting and correcting it takes deliberate testing rather than good intentions.
Reference: Wikipedia: Algorithmic Bias
Related terms
AI Ethics
The study and practice of building and using AI responsibly, covering fairness, privacy, accountability, and harm.
Training Data
The examples a model learns from, whose size, quality, and bias largely determine how the model turns out.
Interpretability
Research into understanding why a model produces the outputs it does, rather than treating it as a black box.