<p>With the advancement of generative AI technology, conversational AI agents (CAIs) have been widely used and shown great development prospects. However, the service failure of CAIs is common and brings negative effects. Therefore, finding ways to mitigate their impacts is crucial. We examine the influence of CAI anthropomorphism on customer forgiveness within service failure based on attribution theory. Specifically, visual and identity cues reduce customer forgiveness, while emotional cues enhance it. Furthermore, in the emotional pathway, visual and identity cues increase anger, while emotional cues reduce anger, and anger reduces customer forgiveness. In the cognitive pathway, emotional cues reduce internal attribution, auditory cues increase internal attribution, and internal attribution reduces customer forgiveness. Finally, the moderating effect of usage scenarios partially holds between anthropomorphism and anger. This study enriches CAI anthropomorphism research and offers guidance for designing more resilient AI services.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Would you forgive anthropomorphic conversational AI agents for service failures? Exploring the impact of anthropomorphism on customer forgiveness

  • Yuguang Xie,
  • Changyong Liang,
  • Peiyu Zhou,
  • Shuping Zhao,
  • Li Jiang

摘要

With the advancement of generative AI technology, conversational AI agents (CAIs) have been widely used and shown great development prospects. However, the service failure of CAIs is common and brings negative effects. Therefore, finding ways to mitigate their impacts is crucial. We examine the influence of CAI anthropomorphism on customer forgiveness within service failure based on attribution theory. Specifically, visual and identity cues reduce customer forgiveness, while emotional cues enhance it. Furthermore, in the emotional pathway, visual and identity cues increase anger, while emotional cues reduce anger, and anger reduces customer forgiveness. In the cognitive pathway, emotional cues reduce internal attribution, auditory cues increase internal attribution, and internal attribution reduces customer forgiveness. Finally, the moderating effect of usage scenarios partially holds between anthropomorphism and anger. This study enriches CAI anthropomorphism research and offers guidance for designing more resilient AI services.