Documentation, Informaiton & Knowledge ›› 2022, Vol. 39 ›› Issue (4): 43-55.doi: 10.13366/j.dik.2022.04.043

Previous Articles     Next Articles

Human-AI Interaction Experience Research: Accelerating the Development of Human-Centered AI

  

  • Online:2022-07-10 Published:2022-10-10

Abstract: [Purpose/Significance] As the two sides of the same coin, "Human-AI interaction"(HAII)and "Human-Centered AI"(HCAI)have become frontier research themes in the field of human-computer interaction and artificial intelligence respectively.Taking the research of HAII experience as the entry point, this study aims to help researchers in such "user-centered" fields as information behavior and user experience consider how to give play to their unique advantages in user study and interactive design, and actively participate in promoting the long-term and stable development of HCAI. [Design/Methodology] This study conducted a literature survey on previous HAII experience research, with an aim to identify the major topics and determine its connections to HCAI research. [Findings/Conclusion] The existing HAII experience research focus on quantifying the influence of human or AI characteristics on the interaction experience. However, the characteristic dimensions of the two subjects were not explored in a systematic manner, and the task and environment components of interaction were not considered. More importantly, the dimensions of user experience investigated were not well connected to the key issues of HCAI. [Originality/Value] A preliminary research system of HAII experience is established: at the level of research design, the interaction subjects, tasks, environment, and experience constitute the core framework; and at the level of research themes, human-AI collaboration, human-AI competition, human-AI conflict, and human-AI symbiosis are promising directions. The research system provides clear guidance for conducting user studies in AI-involved context in the future.

Key words: Human-AI interaction, Interaction experience, User study, Human-Centered AI