I ran an experiment in September. I interviewed an AI chatbot which I had prompted to believe it was President of the United States in 2050. It was a qualified success.
Here is the interview, & my conclusions about its use as a strategic foresight tool.