Even if you take that as an axiom there is still the issue of results being biased (in the academic sense of deviation of expected value from a statistical esti

Even if you take that as an axiom there is still the issue of results being biased (in the academic sense of deviation of expected value from a statistical estimate) related to gender and race. The ethical discussion of the impact of that bias doesn't really change beyond which side of the bias you consider receiving negative vs positive outcomes.