Hm, I thought the overall goal was that you would train LLMs on that data, but the owners of the data would be compensated when output was generated that was in

Hm, I thought the overall goal was that you would train LLMs on that data, but the owners of the data would be compensated when output was generated that was influenced by it. Somehow we have to be able to train LLMs on high-quality information, without having the resulting generative capability destroy the economic support for creating that information in the first place.