Essay.10-FeiyangXie-2100017837

Feiyang Xie

Keywords: XAI; future challenges

Abstract: Proficient language ability serves as a crucial factor distinguishing humans from other species. The primary function of language is to convey information, elucidate phenomena, express ideas, and communicate intentions to others. This communicative process, in turn, fosters the establishment of trust among individuals, as one can clarify their behavior and intentions to others. Despite the remarkable advancements in artificial intelligence driven by deep learning, achieving performance surpassing human capabilities, the critical aspect of interpretability often takes a backseat. When the sole focus is on task performance, the neglect of interpretability hampers the potential for effective human-machine collaboration. The advent of Explainable Artificial Intelligence (XAI) addresses this gap and has emerged as a pivotal research direction. Without the ability to provide understandable and acceptable explanations for its actions, an AI agent faces challenges in gaining human trust, thereby impeding collaboration. In this essay, I will first delve into the challenges that future XAI endeavors are likely to encounter. Subsequently, I will propose the construction of a communicative XAI framework, outlining essential components while acknowledging the limitations borne by each element.

Submission Number: 216