Lessons from the very first chatbot
Summary
David Pierce interviews David Berry and Mark Marino, co-authors of *Inventing Eliza*, about the 1966 chatbot and its creator, Joseph Weizenbaum. They discuss how Eliza’s simple code deceived users into believing they spoke to a human, prompting Weizenbaum’s later warnings about the limits of AI. The conversation connects these historical insights to modern LLMs, exploring the enduring "Eliza effect" and the critical distinction between calculation and human judgment.
Topics discussed
Introduction to Eliza and the new book Inventing Eliza
90 Seconds: Amazon emails, Paramount threat, and Google event
Guest introductions: David Berry and Mark Marino
Joseph Weizenbaum's biography and early computing experiences
The philosophical intent behind creating Eliza
The user experience of interacting with Eliza in the 1960s
Why people believed Eliza was human: The Eliza Effect
Weizenbaum's crisis of faith and Computer Power and Human Reason
The distinction between calculation and human judgment
Comparing the Eliza era to the modern LLM boom
Making AI explainable and breaking the spell of the black box
What Weizenbaum would think of ChatGPT and modern AI
Outro, resources, and subscription information
Listen ad-free on Castria