I've always taken it for granted that discrimination and inequality would be the natural result of any optimization/ML algorithm, especially when maximizing for

I've always taken it for granted that discrimination and inequality would be the natural result of any optimization/ML algorithm, especially when maximizing for profit or any other utilitarian goal. I wonder why people would be surprised by this logical outcome given that it has been long known that statistical discrimination is profitable, and companies rarely have incentives to improve equality and diversity.