The Influence of Assumptions on AI Conversations

A study conducted by researchers from MIT and Arizona State University found that users' prior beliefs about an AI agent significantly influence their interactions and perception of its trustworthiness, empathy, and effectiveness. Priming users with different descriptions of the AI agent influenced their perception and communication with it, even though they were speaking to the same chatbot. Users who were told the AI agent was caring believed it was, and gave it higher performance ratings, while those who believed it was manipulative rated it lower. The study highlights the importance of how AI is presented to society and raises concerns about the potential for deception.