One comment got me thinking about a not-so-distantly related question: Has any one tried get two AI programs talking to each other without either one being told the other was not a human. Would either of them begin to suspect that the other was AI? Would their "conversation" go anywhere interesting or cycle pointlessly?