[skupig]

I'm not really sure what your point is. That was just the most recent paper linked on that repo, which is a convenient list of some relevant papers. There are probably a lot more recent studies, but it does convincingly show that models are still absorbing bias in a way that can affect prediction.

[tptacek]

Again: the papers in the repo don't in fact show that about LLMs (I don't doubt that it could be happening).

[exiguus]

I think the hole root-comment is a joke (if you think about it as training data), because its actually the bias thingy (mensplaining, opportunity vs. knowledge and hn is a very privileged place).