Raw LLM Responses

Finally some recognition that AI cannot be unbiased because it's constantly fed biased data. Even down to simple things like if you have less women in your company then the AI pattern spotting isn't going to just ignore that, it's going to add that to its "What do we want in a candidate" criteria. If you have less women then you have less data for them and that in itself can mean they are looked at less favourably. And there's basic basic stuff like the fact that maybe a company hired a lot less women in the past and those women all tended to leave when they got married or had kids and never come back, and the company has given the system ALL it's hiring data including that historical stuff? Guess what the AI infers from that? Because it doesn't know not to look at the data from 1970-1985 or whatever unless a human thinks to exclude it, and why would they do that if the company they're paying to manage this has said "the more data the better"? And that's assuming that it's not conscious, when we all know that "oh the AI didn't show me any of the CVs with XYZ characteristics oopsy" is just the kind of cover that companies will use to not look at any application where someone disclosed a disability, or where they have marked a certain gender or have names that don't fit English naming conventions, or whatever it is they're filtering out. I'd bet good money that plenty of AI systems have access to the anonymised demographic data that's supposed to be for tracking statistics and trends because hey 'the person doing the hiring still doesn't have access to it so that's okay!' .