Human Connection in the Age of Artificial Intellignce: Perceptions of Empathy, Trust and Authenticity in AI-Mediated Communication

Abstract

The study examines the role of perceived empathy, trust and authenticity related humanness in users' reactions to conversational agents in AI-mediated communication. The study employed a quantitative secondary-data design with survey and interaction data collected from 310 respondents that interacted with an Agent based on the GPT architecture and ELIZA. Participants assessed the empathy, trustworthiness, human-likeness and ability of the agents to establish personal connections. Also studied were the effects of a caring, a neutral and a manipulative motive framing within the GPT condition.The results indicate that the ratings for GPT were significantly higher than ELIZA in empathy, trust, perceived humanness and human connection. Caring motive framing further enhanced empathy, trust, and connection, but did not significantly impact the proxy for authenticity. The results of the correlation analysis indicated that there were strong positive correlations between all focal constructs. The results of the multiple regression analysis revealed that perceived humanness was the most important factor indicating human connection, followed by empathy and trust. After adding these perceptions, agent type was no longer a significant predictor. Results indicated that the extent to which users are related to AI is not merely a matter of the label associated with the technology, but rather, users' sense of relatedness is influenced by how human-like, empathetic and trustworthy they feel while interacting with AI. Transparency, responsible anthropomorphic design, and calibrated trust are emphasized in emotionally sensitive AI applications, in this study.