Model Bias and Fairness
Also written as Algorithmic Bias, AI Fairness, Responsible AI, Disparate Impact
Whether a model's mistakes fall unevenly on particular groups of people, usually because the training data carried a historic pattern. Directly regulated where models touch hiring, lending or insurance.
Think of it like
A hiring filter trained on who got hired before will faithfully reproduce who got hired before.
Junior or senior?
Especially relevant to anything touching recruitment.
Senior sounds like
Has measured error rates group by group and can describe a trade-off they accepted, not just a policy they cite.
Ask them
“How did you check whether the model was wrong more often for some groups than others?”
Sounds like real experience
Describes measuring performance separately by group, and a concrete change or an accepted trade-off that followed from what they found.
Probe further if
Answers with policy language or a vendor's compliance claim, with no measurement they ran themselves.