When I was in middle school (this was in about ’90 or so), I got my hands on an issue of Scientific American from the mid-80s (some Googling says the January ’85 issue). There was an article from AK Dewdney, called “Artificial Insanity: When a Schizophrenic Program Meets a Computerized Analyst,” which was about Racter and Joseph Weizenbaum’s Eliza, the Rogerian therapist chatbot from the early 1960s. One of the most fascinating bits was when Eliza began an analysis of Racter; both the brittleness of Eliza’s response engine, and flimsiness of Racter’s flights of fancy, were made clear.
I remember being fascinated by this at a young age, though, in spite (or maybe because!) of the strangeness of computers talking to each other, and that their attempts to seem human fell short relatively quickly.