Understanding Technology Acceptance of Social Robots as Conversational Interfaces for LLMs

Abstract

Advancements in large language models (LLMs) have transformed human-computer interaction, enhancing engagement with conversational agents like chatbots, virtual agents, and social robots. This study explored how different conversational interfaces (computer, Furhat, NAO, and Pepper) impact perception and acceptance using the Human-Robot Interaction Evaluation Scale (HRIES) and the Technology Acceptance Model (TAM). One hundred participants rated interfaces through a video-based survey, revealing significant differences: computers scored highest in Perceived Usefulness (PU) and Ease of Use (PEU), while social robots, particularly Pepper and NAO, excelled in Sociability. A Structural Equation Model (SEM) indicated that HRIES dimensions of Sociability and Agency positively influenced Perceived Enjoyment (PENJ) and PEU, whereas Disturbance negatively affected both. These findings highlight the nuanced interplay between interface design and user acceptance, suggesting that while factors like Sociability and Agency are pivotal for enhancing enjoyment and ease of use, they alone are insufficient to account for overall acceptance.