Skew in a model's output that traces back to skew in its training data or design choices.
Why it mattersIt shows up quietly - in who gets shortlisted, whose accent transcribes badly, which photo it picks.
See also Alignment, Explainability
Skew in a model's output that traces back to skew in its training data or design choices.
Skew in a model's output that traces back to skew in its training data or design choices.
Why it mattersIt shows up quietly - in who gets shortlisted, whose accent transcribes badly, which photo it picks.
See also Alignment, Explainability
Browse the full glossary for plain-English definitions across models, agents, data, and safety.
Back to all terms