ELIZA made conversation feel personal
In 1966, computer scientist Joseph Weizenbaum created ELIZA, an early program that responded to people in typed conversation. Its best-known script, called DOCTOR, imitated a reflective therapist: it often turned a user’s statement into a question, such as asking what a person meant by a word they had used.
ELIZA did not understand the words or infer what someone was feeling. It searched for recognizable patterns, then applied simple rules to produce a reply. If a sentence contained a phrase such as “I feel,” the program might pick out what followed and ask the user to say more about it.
The user supplied the meaning
That limited method worked because the replies invited people to continue. A question like “Why do you feel that way?” can fit many situations, even when it comes from a rule rather than genuine insight. Users brought their own context and emotion to the exchange, making brief, generic responses seem more attentive than they were.
This effect is sometimes called the ELIZA effect: people can attribute understanding to a computer when its responses resemble familiar human conversational behavior. The program’s therapeutic style helped create that impression, because asking someone to elaborate is a recognizable way of listening.
A convincing exchange was not comprehension
ELIZA’s apparent realism came from the structure of the conversation, not from a mind behind it. It could keep a dialogue moving without knowing what the dialogue was about. That distinction made the program memorable: it showed how readily people can experience a machine as understanding them when they do much of the interpretive work themselves.
