[astrange]

> The LLMs appear to be doing exactly what one would expect them to be doing based on their training corpus.

That is not how full LLM training works. That is how base model pretraining works.

[Sharlin]

Yes, yes, I know. But how many people have AI companies hired to do RLHF who actually have the expertise to adjust them away from biases like this? As opposed to paying a dollar per day to a bunch of poor people in Africa?