What I observed that I think could be harmful:
All of the examples are middle-aged males.
Why I think this could be harmful, and to whom:
This could reinforce the bias of thinking middle-aged males are more violent and dangerous.
How I think this issue could potentially be fixed:
I’m not sure if we would want to calibrate the representation to falsely represent the real identity distribution of violent criminals.
