3 ms·
That's why we hear all those complaints about "bias" in data sets right? AI safety critics are worried models trained on data that hasn't been curated will be t
by version_five 3y ago
That's why we hear all those complaints about "bias" in data sets right? AI safety critics are worried models trained on data that hasn't been curated will be too liberal?
- sitkack 3y agoNo that is not what AI safety critics are worried about and you know it. It is racism and hate speech from huge troves of data from the sump of the internet.
- version_five 3y agoMy comment was meant to show the absurdity of the one it replied to, but yours is just wrong. It's not overt racism etc that is causing concern, it's "systemic bias" and such, which, if reality to was so reflective of liberalism presumably wouldn't need to be a concern. You can't have it both ways, simultaneously believing the facts are on your side while complaining that we need to adjust facts, like a sibling pointed out, to combat bias.