If we had some way of scoring candidates numerically, then yes, you'd just pick the biggest number. But I've never heard of a hiring system that worked like that. There's always some subjectivity. You can look at a bunch of resumes and rank them, but a good chunk of that ranking is guesswork and opinion. That gives room for your biases to play, and you'll end up with the "best" candidates tending to match those biases. And don't tell me you don't have any racial or gender biases; you do, everybody does.
You shouldn't give up a 9.9/10 because he's white, and hire a 2.4/10 because she's black. But if you have a bunch of candidates around 8/10, consider hiring the minority candidate who's a 7.9 rather than the candidate who matches the existing demographics of your team and is an 8.1. Your numbers are probably ±3 anyway, so it's not the irrational decision it sounds like it would be in a universe of pure numbers.