The enshittification of LLMs began when chat models started asking leading questions at the end of their response to prompt you to continue the conversation