>> One problem with AI (vs e.g. Google search) is that there is no mechanism for attribution. Sorry if this is a silly question, but is that really the case? C

>> One problem with AI (vs e.g. Google search) is that there is no mechanism for attribution. Sorry if this is a silly question, but is that really the case? Can we not train an LLM on successively larger training sets, each will uniquely ID's model and associated better performance?