I keep my past conversations around as a sort of project note. As soon as GPT gets in a pattern of giving broken answers or broken code it will basically never come back and be able to correct itself. Do to this, I abandon conversations often and start a new one where the previous GPT started messing up. As of late I’m finding that a new conversation does nothing to alleviate the issue because GPT constantly references our past conversations and just keeps bringing its past, incorrect answers to the new conversation. Before someone goes there, the solution is absolutely NOT to delete past conversations. It should NOT be a requirement that I delete my history in order to get GPT to give a decent answer. What is being done to correct this situation? It has be paralyzed.
Thanks