You can teach the ML to find bad applicants quicker because this is much easier to do
- the HR person usually has gut reactions and reasons for why denying and we can find these
- its harder to do the more grey area of deciding based on more factors
How can we pull out the reasons the AI does what it does and highlight them?
- Humans need to TRUST machine and see reasons
Small Factors that are underrepresented in training dataset are continued to be underrepresented
- Year in school...if grad before 2010...much lower chance!
- AI finds patterns that DO REALLY exist and highlights them
When using previous hires as training sets, certain populations are more represented