Anthropomorphism and customers’ willingness to use artificial intelligence service agents

Previous studies provide inconsistent evidence regarding the effect of anthropomorphism on customers' willingness to use AI service agents. This paper explains the reason by introducing service context. Two situational experiments are used to demonstrate that under the context of high perceived control, customers expect AI service agents with more anthropomorphic designs to perform better and prefer highly human-like AI service agents. However, under the context of low perceived control, customers perceive stronger threat in facing AI service agents with more anthropomorphic designs and prefer less human-like AI service agents. Moreover, we find that this effect is significant only in social scenarios. These findings provide new insights into previous inconsistent evidence regarding anthropomorphic design's influence on customers' willingness to use AI service agents. Our findings also have important implications for AI service agents design in different service contexts and advance the literature on human–robot interaction and marketing.