My Conversation With ELIZA
The author recounts a recent interaction with a web-based recreation of ELIZA, the early AI chatbot, and shares a transcript of the stilted conversation. He expresses skepticism about ELIZA's historical reputation and criticizes anyone who found it useful as a virtual therapist, calling such people "suffered-a-permanent-head-injury wrong."
Background
- ELIZA was one of the first chatbot programs, created in the 1960s by MIT's Joseph Weizenbaum. Its DOCTOR script mimicked a Rogerian therapist using simple pattern-matching: it transformed user statements into questions (e.g., "I feel X" → "Why do you feel X?").
- Weizenbaum was shocked when people — including his own secretary — became emotionally attached to ELIZA, treating it as a real therapist. This became a famous early example of humans anthropomorphizing computers.
- The author is trying a modern restoration by the ELIZA Archaeology Team, which recovered original 1960s source code. The transcript reveals how primitive the program actually is (grammar errors, inability to handle basic negation).
- This matters because today's AI chatbots (ChatGPT, etc.) have revived debates about human attachment to machines. The author argues that if people got attached to something as brittle as ELIZA, that's far stranger than relying on modern chatbots — pushing back on the romanticized history of ELIZA.