论智慧税务中的法技价值对齐  

On the Value Alignment between Law and Technology In Smart Taxation

在线阅读下载全文

作  者:许多奇 董家杰 Xu Duoqi;Dong Jiajie(Fudan University)

机构地区:[1]复旦大学法学院,上海200438

出  处:《武汉大学学报(哲学社会科学版)》2025年第3期28-38,共11页Wuhan University Journal:Philosophy & Social Science

基  金:国家社会科学基金重大项目(22&ZD203)。

摘  要:在发展新质生产力的背景下,人工智能赋能智慧税务需要防范对税法价值取向和基本原则的偏离。人工智能对齐问题源自技术范式演进带来的自主性跃升,本质上是一种基于目标意图的价值对齐,在税收征管智能化进程中表现为人工智能技术与税收征管法治之间的张力调和。从智慧税务的具体应用场景来看,税务人工智能的权力寻求倾向、代理不适格以及内外部归责缺位冲击征管法治原则,泛化不足和偏见循环则将给征管公平造成挑战,征管效率的实现也可能因人工智能幻觉的存在而受到阻滞。对此,既应立足数据优化和“人在回路”强化技术应对,也要探求权利保障和义务前置的法治路径,并在包容审慎理念下规避“技术解决主义”陷阱。In the context of developing new quality productive forces,AI-enabled smart taxation needs to prevent deviations from the value orientations and basic principles of tax law.The issue of AI alignment,stemming from the enhanced AI autonomy brought about by the evolution of technological paradigms,is essentially a kind of value alignment based on objectives and intentions,which is manifested as the tension reconciliation between AI technology and the rule of law in the smart transformation of tax collection and administration.From the perspective of specific application scenarios of smart taxation,the power-seeking tendency,agency ineligibility and absence of internal and external accountability of AI will impact the principle of the rule of law in tax administration,while the insufficiency of generalization ability and the feedback loop of bias will challenge fairness and the realization of the efficiency may be hindered due to hallucinations about AI.In this regard,it is necessary to not only strengthen the technical response based on data optimization and human-in-the-loop approaches,but also explore the path of the rule of law that protects rights and prioritizes obligations,and avoid the trap of technosolutionism with an inclusive and prudent approach.

关 键 词:智慧税务 税收征管 人工智能 税法原则 价值对齐 纳税人权利 算法决策 算法治理 

分 类 号:D912.29[政治法律—民商法学]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象