Data & Machine Learning
Responsible AI, Bias & Fairness
8 practice questions. Free questions open a full answer guide; the rest unlock with Pro.
- You ship a model and notice it performs noticeably worse for one underrepresented group than for everyone else. What could be causing that, and how would you start addressing it?
- A stakeholder wants the model to be 'fair across all groups' and satisfy every standard fairness metric at once. How do you explain what's actually achievable, and how do you decide which fairness criterion to optimize?Go Pro
- You've found that a deployed model disadvantages a protected group. Where in the pipeline would you intervene to mitigate the bias, and what are the trade-offs of pre-processing, in-processing, and post-processing approaches?Go Pro
- Where does bias actually come from in a machine learning model? Walk me through the main sources.Go Pro
- Your model looks fair on the top-level group metrics, but you suspect it's failing specific subgroups. How would you detect and handle intersectional or subgroup fairness problems?Go Pro
- How would you go about detecting whether a deployed model is treating some demographic group unfairly, and how is that different from just looking at overall accuracy?Go Pro
- Walk me through the common fairness metrics — like demographic parity, equalized odds, and calibration. Why can't a model usually satisfy all of them at once, and how would you choose which one to optimize for?Go Pro
- You removed gender and race from your training data, but the model is still producing biased outcomes across those groups. How is that possible, and what would you actually do about it?Go Pro
Want questions matched to your role? Paste a job title, job description, or CV for a personalized set, or go Pro to unlock the full bank.