The fact that this got proven to be unfair at all and convictions overturned etc. is because it could ultimately be pointed out that these specific software bugs existed in an intentionally designed piece of software. But it still took years to get past the "computer says no" attitude to being told they were wrong. I worry as increased use of AI and profiling becomes normal, that there will be no avenue for people when e.g. the background check company's ML system says they're a high risk candidate.