Not inherently, but things can be tested.
If you have a bunch of otherwise identical résumés, with the only difference being the racial connotation of the name, and the AI gives significantly different results, there's an identifiable problem.
Not inherently, but things can be tested.
If you have a bunch of otherwise identical résumés, with the only difference being the racial connotation of the name, and the AI gives significantly different results, there's an identifiable problem.