The mind in the machine: Estimating mind perception's effect on user satisfaction with voice-based conversational agents

Abstract

Given the interest in imbuing voice-based conversational agents with humanlike features, understanding how this affects user satisfaction is ultimately important for business performance. Mind perception theory explains how ascribing the mental capacity for agency and experience to artificial intelligence shapes subsequent user attitudes. Hence, we estimate the effect of mind perception on satisfaction in users with high/low innovativeness using data from text-based online reviews, which better reflect actual usage than traditional surveys. Methodologically, where numerous controls affect the cause and outcome variables in a model, traditional machine learning methods produce biased estimates. We overcome this by deploying Double/Debiased Machine Learning (combined with text analytics). Results show that user satisfaction is increased by two forms of perceived experience: directed at moral agents, or moral patients. Perceived agency, however, has no significant influence. The increase in satisfaction from both types of perceived experience is stronger among users with high (vs. low) innovativeness.