I'm curious why and how models like these give one answer for one person and a completely different answer for someone else. One reason can be memory maybe? Pas

I'm curious why and how models like these give one answer for one person and a completely different answer for someone else. One reason can be memory maybe? Past conversations that tell the model "Think this way for this user"