Article Open Access http://dx.doi.org/10.26855/acc.2025.04.006
Research on Methods for Improving the Explainability of Artificial Intelligence Based on Causal Reasoning
Lingling He
1Zhejiang Ronghe Tea Culture Co., Ltd., Hangzhou 310000, Zhejiang, China.
2Yunnan Ronghehao Tea Industry Co., Ltd., Kunming 650000, Yunnan, China.
3Kunming Zhonggu Tea Industry Co., Ltd., Kunming 650506, Yunnan, China.
*Corresponding author: Lingling He
Published: May 13,2025
Abstract
In the era of rapid development of AI, although AI systems have made significant progress in performance in many areas, the lack of interpretive capability has been a major obstacle to further popularization and application, especially in key areas such as medicine, finance, and autonomous driving. Causal reasoning aims to study the cause-and-effect relationships between variables, which gives a new look at improving the interpretability of AI. This article discusses ways and means to increase the interpretability of AI through causal reasoning. Initially, the meaning of the respective concepts and their connection with the interpretability of AI and other theoretical foundations is laid out. After that, concrete paths of realization are given, such as cause-and-effect modeling, cause-and-effect analysis, and generation of interpretation of cause-and-effect instructions. Testing specific examples, the effectiveness and practical value of these methods are confirmed in typical use cases. Integrating cause-and-effect connections into an AI system can make the AI decision-making process more transparent and understandable, which is of great importance for improving AI reliability and recognition.
References
[1] Zhang Z. Research on the prediction method of judicial judgments based on causal analysis [dissertation]. Shanxi: Shanxi University; 2023.
[2] Cai Y. From Bayesian inference to causal inference: a study on Bayesian classification models based on causal networks [dissertation]. Jilin: Jilin University; 2024.
[3] Li M. Research on the knowledge expression of traditional Chinese medicine practitioners' differential diagnosis and treatment of chronic heart failure based on dynamic uncertain causal graph [dissertation]. Shandong: Shandong University of Traditional Chinese Medicine; 2023.
[4] Guo Z. A study on interpretative auxiliary diagnosis and causal effect estimation methods [dissertation]. Beijing: Beijing Jiaotong University; 2022.
[5] Sun Y. Research on transfer reinforcement learning control method based on causal modeling [dissertation]. Jiangsu: Southeast University; 2023.
[6] Bao C. Lightweight cervical TCT pathological image cell detection model and interpretability study [dissertation]. Shanghai: Donghua University; 2023.
[7] Moraes DAI, Arrighi L, Junior BS, et al. Explainable artificial intelligence (xAI) applied to deep computer vision of microscopy imaging and spectroscopy for assessment of oleogel stability over storage. J Food Eng. 2025;394:112515.
[8] Kruk M. SHAP-NET, a network based on Shapley values as a new tool to improve the explainability of the XGBoost-SHAP model for the problem of water quality. Environ Model Softw. 2025;188:106403.
How to cite this paper
Research on Methods for Improving the Explainability of Artificial Intelligence Based on Causal Reasoning
How to cite this paper: Lingling He. (2025) Research on Methods for Improving the Explainability of Artificial Intelligence Based on Causal Reasoning. Advances in Computer and Communication, 6(2), 87-93.
DOI: http://dx.doi.org/10.26855/acc.2025.04.006