Influencing driving safety by matching AI assistant's verbal emotions to driver: A randomized controlled trial on performance, attention, and emotion

Abstract

This Research explores the emotional alignment and conflict between AI assistants and drivers, and their impact on the driver's emotional state, behavior, and attention. As intelligent connected vehicles evolve, voice interaction has become a key medium for emotional communication. Previous studies highlight the influence of driver emotions on driving performance, with emotional mismatches potentially exacerbating fluctuations and impairing performance. To investigate this, a multimodal emotion induction experiment was conducted, where drivers were induced into pleasant and frustrated emotional states and paired with AI assistants expressing either positive or negative emotions. A simulated driving test collected data on driving performance, eye movements, and electrodermal activity (EDA). Results showed that positive emotions enhanced driver attention, decision-making, emergency response, and reduced errors and accidents. When emotionally consistent, AI assistants improved driver performance; however, emotionally conflicting assistants triggered self-regulation mechanisms in frustrated drivers, enhancing attention and response speed. These findings provide a theoretical basis for using emotional guidance and human-machine collaboration strategies to improve driver performance and traffic safety.