A hiring model favors one group because of historical data. What is this an example of?
Why C is the answer
Bias in training data
Models learn patterns in historical data, including unfair ones. Trustworthy AI work is about detecting and reducing this.
In detail
- Patterns copied from historical data are bias in training data.
- Removing a sensitive column may not remove the bias.
Use case
An illustrative scenario, not a real incident.
Past hiring favored one group, and a model trained on it repeats that pattern.
See it in different ways
flowchart LR A["Consent for data"] --> B["Balanced training data"] B --> C["Test across groups"] C --> D["Guardrails in production"] D --> E["Monitor and explain"]
mindmap
root((Trustworthy AI))
Consent
Before collecting data
Fairness
Test across groups
Transparency
Say that it is AI
Safety
NeMo Guardrailsflowchart LR S0["Biased history"] --> S1["Training data"] --> S2["Model"] --> S3["Unfair outcomes"]
Key point: Models learn unfair patterns from history.
Why the other options are wrong
- A. Quantization error
- Quantization is about number precision.
- B. Tokenization drift
- That is not the issue here.
- D. Overfitting
- Overfitting is about generalizing, not fairness.