OK.

Can we suspend identity politics and admit that systemic homogeneity bias in software development is as much a technical as social and political problem?

I mean, why is it ok to release facial recognition software that has only been trained on white faces? Why is that considered a minimum viable product?

When we get to the point that facial-recognition systems don't recognize black people, autonomous cars won't even stop for them and law enforcement risk assessment AIs automatically tag them with a x2 modifier for criminality, this stuff stops looking like naive techie ignorance and starts looking like malice.

And, all else aside, it's bad design.