AI Ethics & Bias

    AI · Beginner

    Definition

    AI ethics studies fairness, privacy, transparency and safety of AI systems.

    Biased data produces biased decisions that affect real people.

    Diagram

    Data→ Model→ Decision→ Impact

    Flowchart

    Audit training dataMeasure fairness metricsExplain predictionsMonitor in production

    Example: Selection-rate ratio

    rate_a, rate_b = 0.6, 0.45
print(round(rate_b / rate_a, 2))

    Output: 0.75

    Key points

    • Ratio below 0.8 often flags disparate impact
    • Keep humans in the loop for high-stakes decisions

    Interview questions

    Where does bias come from?

    Unrepresentative data, biased labels and proxy features.

    More AI topics

    What is Artificial IntelligenceNeural NetworksGradient DescentCNNs & Computer Vision