“Even if these sources of data bias could be identified and corrected, however, there may still be some group-based differences that are not attributable to data bias. If so, groups may experience different risk scores and categories that would not necessarily indicate bias. Further, it is often difficult (or impossible) to discern whether some observed group-level differences in data are genuine or reflect some sort of systemic bias.“
Further, it is often difficult (or impossible) to discern whether some observed group-level differences in data are genuine or reflect some sort of systemic bias.
The fact that systemic bias is sufficient to make someone say this is kind of problematic, no?
Hopefully it’s clear this is super problematic in the context of risk assessment tools. It means if you are categorized as part of a racial group that is convicted of more crimes, you might be categorized as a higher risk than you would otherwise.
They argue models from education and psychology that address racial disparities could be adapted, but those institutions are designed to help people rather than to decide if they should be in jail.
Just in case people aren’t familiar, some context from the aclu.
Also note: the is regarding the united states