Fact is, the only reason why openAI has become usable is for #mainvolume proving it can never become causality aware regardless how large the models are and how much training it takes, making it safe for make believe, well at least not without violating intellectual properties to get thieving.