AI Fortune-Teller: Juxtaposing Shaman and AI to Reveal Human Agency in the Age of AI

TL;DR

This paper explores how people perceive AI advice by comparing it to traditional shamanic fortune-telling, revealing that perceptions of advice are unaffected by its source once initially formed.

Contribution

It demonstrates that people's attitudes toward AI advice remain stable even after revealing the advice's traditional, non-AI origin, highlighting perceptions of human agency and trust.

Findings

Participants' attitudes did not change after learning the advice was from a shaman.

Perceived importance of AI's explainability may be overestimated.

Human perceptions of advice are resilient to source disclosure.

Abstract

This speculative video piece showcases participants interacting with a career counseling AI agent, unaware that the responses were actually derived from the fortunetelling of a mudang (a Korean traditional shaman). Our work captures this deception and documents participants' reactions, showcasing shifts in their initial perceptions of the agent's advice following the reveal. Notably, even after learning that the advice came from a mudang rather than an AI, participants did not change their initial attitudes toward the advice they received. This raises questions about the perceived importance of AI's explainability and accuracy. By juxtaposing scientific and pre-scientific approaches, we aim to provoke discussions on human agency in the age of AI. We argue that, regardless of AI's advancements, we continue to navigate life in fundamentally human ways -- wonderfully messy and uncertain.