Fairness metrics, algorithmic bias, explainability, privacy, alignment, the EU AI Act and responsible AI frameworks.
40 cards · basic cards · AI-written, checked twice. Edit anything.
- What does demographic parity require in algorithmic fairness?
Show answer
Model predictions should be independent of protected attributes or demographic groups- Define equalized odds as a fairness metric.
Show answer
Equal true positive and false positive rates across different demographic groups- What is individual fairness in AI?
Show answer
Similar individuals should receive similar treatment from an algorithm regardless of group membership- What does calibration mean in fairness contexts?
Show answer
Predicted probabilities accurately reflect actual outcomes across all demographic groups- What is selection bias in machine learning?
Show answer
Bias introduced when training data is not representative due to non-random sampling or data collection- Define representation bias.
Show answer
Training data underrepresents certain demographic groups or populations- What is measurement bias?
Show answer
Systematic error in how variables are measured, defined, or collected across groups- Explain feedback loops in algorithmic bias.
Show answer
Past biased predictions influence future training data, perpetuating and amplifying the original bias- What is historical bias in AI?
Show answer
Bias reflecting historical inequalities or prejudices present in the world or training data- What is the fairness-accuracy tradeoff?
Show answer
Optimizing for fairness often requires accepting lower overall model accuracy or performance- What is explainable AI (XAI)?
Show answer
Making AI and machine learning model decisions transparent and understandable to humans- What does LIME stand for?
Show answer
Local Interpretable Model-agnostic Explanations, explains predictions by approximating model behavior locally- What is SHAP?
Show answer
SHapley Additive exPlanations, assigns importance values to features based on cooperative game theory- What is the difference between interpretability and explainability?
Show answer
Interpretability is inherent transparency in model design; explainability is post-hoc explanation of decisions- What are black box models?
Show answer
Models whose internal decision-making process is opaque, complex, or difficult for humans to understand
FlashKeepers