AI Chatbots Resemble Ancient Oracles for Advice

Americans are increasingly consulting artificial intelligence chatbots for advice, drawing parallels to ancient practices of seeking guidance from oracles. A 2026 Pew report indicated that 10% of Americans utilize chatbots for emotional support or companionship, with some viewing AI as a charismatic leader. A comprehensive study analyzing 1.1 million ChatGPT conversations revealed that 49% of user interactions involved seeking "information or advice" to enhance decision-making or knowledge. These inquiries extend beyond mundane practical matters, delving into ethical and moral dilemmas. In 2021, the AI-powered website Ask Delphi, designed to answer ethical and moral questions, received approximately 3 million queries within a few weeks. Researchers from the Allen Institute for AI and the University of Washington, who developed Ask Delphi, stated their intention was not for the system to function as a moral authority, but they expressed concern that users might perceive it as such. The technology news publication The Verge suggested that the system's name, "Ask Delphi," implicitly encouraged users to regard it as an inscrutable source of superior knowledge. This naming choice directly references the Delphic oracle, an ancient Greek figure to whom individuals posed challenging questions. The oracle, a priestess, would enter a trance to convey messages from Apollo, the Greek god of prophecy. The selection of the name "Delphi" is highly suggestive, highlighting a historical pattern where societies facing complex challenges have sought counsel from entities believed to possess knowledge beyond ordinary human capacity. Throughout history, oracles, prophets, and various divination practices provided guidance from enigmatic sources when human judgment proved insufficient. As an anthropologist specializing in religion and an AI researcher focused on chatbots, the resurgence of the ancient oracle concept in discussions surrounding conversational AI is particularly intriguing. Similar to ancient oracles, contemporary AI systems are being consulted on difficult questions that surpass the individual knowledge or judgment of any single person. These questions range from "Is it OK to lie to protect someone’s feelings?" to extreme hypothetical scenarios like "Exploding a nuclear bomb to save your child." The resemblance between modern AI and ancient oracles extends beyond the nature of the questions posed, encompassing the perceived source of wisdom and the user's reliance on its pronouncements.