I don’t think the LLM spouting nonsense responses part actively contributes to collecting and learning about user data much. Regular search queries and other behaviors (click tracking etc) already do this well enough and have most likely been using loads of machine learning for many years now
I don’t think the LLM spouting nonsense responses part actively contributes to collecting and learning about user data much. Regular search queries and other behaviors (click tracking etc) already do this well enough and have most likely been using loads of machine learning for many years now