Essay.10-WeinanQian-2100017831
Keywords: Explainable artificial intelligence, Challenge, Explainability, Evaluation
Abstract: With the advancement of artificial intelligence (AI) technology, the demand for more explainable AI systems has increased. There is a growing curiosity to understand how AI models make decisions and produce outputs. Explainable Artificial Intelligence (XAI) aims to develop AI systems capable of explaining their behavior while maintaining high performance in specific tasks. However, XAI is still in its developmental stages and specialists in this field face numerous challenges. In XAI, researchers must balance the trade-off between system explainability and performance. Quantitative metrics to evaluate XAI's explainability are lacking, and scientists primarily rely on costly and subjective human experiments to assess its effects. This paper delves into these two key challenges in XAI and proposes potential solutions, leveraging an example of XAI application in robotics to illustrate these challenges.
Submission Number: 224
Loading