I wonder if this is algorithm bias or teaching material bias? I can't see anyone writing "if name not in white_male_names then score -=15", but take successful

I wonder if this is algorithm bias or teaching material bias? I can't see anyone writing "if name not in white_male_names then score -=15", but take successful profiles, generalize and create predictive model. Anyways, this should be pretty easy to test: Just rename applicants with pseudonyms before rating.