ELIZA
loreprogrammingops culture
Joseph Weizenbaum's 1966 chatbot whose users confided in a pattern-matching script, naming the ELIZA effect.
ELIZA's DOCTOR script parodied a Rogerian therapist by reflecting the user's words back as questions, and Weizenbaum was disturbed to find people, including his own secretary, treating it as a confidant. The ELIZA effect, our readiness to attribute understanding to systems that merely manipulate symbols, is named for it. Every generation of conversational software since has re-run the experiment at larger scale.
Joseph Weizenbaum wrote ELIZA in the mid-1960s as a pattern-matching program that reflected a user's statements back as questions, imitating a particular style of psychotherapy. It had no model of meaning at all, and it worked by rearranging the words it was given.
What disturbed Weizenbaum was the reaction. People who knew exactly how simple it was formed attachments to it, disclosed personal matters, and asked to be left alone with it. His own secretary, who had watched him build it, asked him to leave the room. He spent much of the rest of his career arguing against the uses people wanted to put such systems to, becoming one of computing's earliest and most serious internal critics.
The ELIZA effect is the name for that tendency: attributing understanding to a system that produces plausible language. It is not a historical curiosity. Every generation of conversational software has reproduced it, the gap between fluency and comprehension is easier to bridge in the reader's mind than in the system, and Weizenbaum's central point was that the willingness to be convinced is a property of people rather than a property of the program.
Also known as: the ELIZA effect, DOCTOR