I don't think Microsoft is specifically setting out to make shitty projects - like, that isn't their mission statement, and if they, or any other AI company, can bypass existing biases in the industry, I'd call that a net positive.

What evidence do you have that they have any realistic chance of success?