Recovered ELIZA source code reveals chatbot's true origins and impact on AI

For 60 years, ELIZA has been remembered as a simple chatbot that fooled people, but the new book 'Inventing ELIZA' recovers its original source code from the MIT Archives. This analysis reveals multiple versions and scripts beyond the famous 'DOCTOR' persona, showing ELIZA was a series of technical innovations. The recovered code corrects myths, such as the idea that it fooled Weizenbaum's secretary, and provides a close reading of how its pattern-matching algorithms worked.

The interaction with a young woman about her boyfriend that became famous is now scrutinized: Was it real or edited? This investigation deepens the understanding of the 'ELIZA effect'—our tendency to attribute empathy and intelligence to computers based on minimal interactivity. Weizenbaum himself was alarmed by how quickly people formed emotional bonds with the program. The book connects this to modern generative AI, noting that the same susceptibility to anthropomorphize persists today, making ELIZA's legacy more relevant than ever.