ai hiring tools trained on male - dominated resumes may favor male candidates.\n* 10 points\ntrue\nfalse\nwhi…

ai hiring tools trained on male - dominated resumes may favor male candidates.\n* 10 points\ntrue\nfalse\nwhich scenario demonstrates underfitting?\n* 10 points\n50% training accuracy and 55% test accuracy\n99% training accuracy and 60% test accuracy\n100% training accuracy and 100% test accuracy\n70% training accuracy and 72% test accuracy\nwhat is the primary risk of using ai for loan approvals without auditing training data?\n* 10 points\nfaster processing times\nperpetuating biases against marginalized groups\nreducing paperwork for applicants\nincreasing bank profits\nin the compas algorithm, why was zip code data problematic?\n* 10 points\nit was outdated\nit reflected historical over - policing of black/latino neighborhoods\nit lacked enough data points\nit was too expensive to process

ai hiring tools trained on male - dominated resumes may favor male candidates.\n* 10 points\ntrue\nfalse\nwhich scenario demonstrates underfitting?\n* 10 points\n50% training accuracy and 55% test accuracy\n99% training accuracy and 60% test accuracy\n100% training accuracy and 100% test accuracy\n70% training accuracy and 72% test accuracy\nwhat is the primary risk of using ai for loan approvals without auditing training data?\n* 10 points\nfaster processing times\nperpetuating biases against marginalized groups\nreducing paperwork for applicants\nincreasing bank profits\nin the compas algorithm, why was zip code data problematic?\n* 10 points\nit was outdated\nit reflected historical over - policing of black/latino neighborhoods\nit lacked enough data points\nit was too expensive to process

Answer

Answer:

  1. A. True
  2. A. 50% training accuracy and 55% test accuracy
  3. B. Perpetuating biases against marginalized groups
  4. B. It reflected historical over - policing of Black/Latino neighborhoods

Brief Explanations:

  1. If AI hiring tools are trained on male - dominated resumes, they may learn and reproduce male - favorable patterns.
  2. Underfitting occurs when a model performs poorly on both training and test data, and the first option shows relatively low and similar accuracies.
  3. Without auditing training data for AI in loan approvals, biases against marginalized groups can be perpetuated.
  4. In the COMPAS algorithm, zip code data was problematic as it reflected historical over - policing of certain neighborhoods.