Human After All

Abstract

In automated cars social agents could facilitate trust generation. We conducted a driving simulator study in which we manipulated the human-likeness of a spoken-dialogue system (SDS). In a simple rural road overtaking paradigm users either interacted with a SDS with no physical presence (Voice Only), with an SDS represented by the humanoid robot NAO providing road advice (NAO) or with a NAO with additional social interaction (Social NAO). The Social NAO condition showed a significantly higher trust as compared to the Voice Only condition. All group differences were associated with medium effect sizes. This study provides some evidence that anthropomorphic features increase trust in automated driving systems.