The whole concept should be simple: if the people collecting the data are selling it to "AI" companies for money, then the person or people who created those data, which takes much more effort than just collecting it, should also be fairly compensated. These ideas that they should "get over it" or "move on" are nonsensical, to say the least.
Furthermore, the "AI" models should be designed from the outset to provide citations, much like any credible researcher would, indicating where the source data came from. That, in turn, would make it much easier to determine who is entitled to royalty payments.