A current example is using an AI to determine whether a set of parents is viable to continue being parents or if they should have custody taken away from them. Most states don’t allow their use but some do, and if they have metrics that include permanent features that can’t be changed, such as IQ, etc, it becomes a discriminatory tool whereby something like a single visit by social services is enough to tip the scales into “remove the children” status. A case in PA going on right now is what brought it to mind
[responsivevoice_button voice="US English Male"]