Explainable AI: A Way to Achieve Trustworthy AI

Explainable AI: A Way to Achieve Trustworthy AI

연구 분야: Artificial Intelligence

학회: 2024 IEEE 10th Conference on Big Data Security on Cloud (BigDataSecurity)

AI is black-box and non-explainable, in other words, due to the complexity of the decision-making process of AI, people are unable to know why and how AI makes the decision. For these reasons, people will question and worry about AI’s decision-making. Against this background, researchers have proposed the concept of Explainable Artificial Intelligence (XAI). The aim of this paper is to help readers be able to better understand XAI through extensive surveys. In this paper, we discuss the concept of XAI, the motivation of the research, the current status of the research, and the difficulties in the research process. At the same time, we also discuss the specific methods to realize AI explainability by dividing AI explainability methods into two ways: transparent design and ex post facto interpretation. In addition, we select the three fields of healthcare, finance, and autonomous driving, discuss the application of XAI in these three fields, and select some representative results as examples through surveys. Finally, we summarize this article and present our outlook on the future development of XAI.