[Gustoo]

To be more specific. Lets say I make a spreadsheet that calculates risk.

5% risk associated with location A (based on high school graduation rates of the people there)
15% risk associated with location B (based on high school graduation rates of the people there)

and other risk factors that I assign a value to with a math tool (a spreadsheet). This risk % is calculated equally based on the given rates of graduation, or whatever the variable.

I put your info into a few cells in a spreadsheet, and my tool says 72% safe, and my minimum safe percent is 75% , so I reject you.

Did my tool (me) reject you, or did "math" reject you.

I am making a tool like this right now to support a business decision. Depending on how I present the information I can make the decision as positive or negative as possible. When discussing these things for the collective good of my organization I take pains to explain the riskiest possible perspective, and the least risky possible perspective using the same data.

I can easily make a tool that applies equal "fair" weights to all the variables that will provide rejection status for whatever particular group or area that I want to be rejected. I can cater the tool to my needs. The tool objectively uses provided data and fairly applies weights but always rejects certain types of people. I can make the tool do what I want. There is an error in assuming the risk assessment tools we create for ourselves are somehow neutral. We make them. We don't have to listen to what they say and as a society we should have a say in the way these things are handled. Math isn't racist, but my math based tool can easily be.