This study investigates the ethical grounding of the European Union Artificial Intelligence (EU AI) Act by using Semantic Textual Similarity (STS) to analyze the alignment between normative ethical theories and regulatory language. Despite being regarded as a significant step toward regulating Artificial Intelligence (AI) systems and its emphasis on fundamental rights, the EU AI Act is not immune to moral criticism regarding its ethical foundations. Our work examines the impact of three major normative theories of ethics, virtue ethics, deontological ethics, and consequentialism, on the EU AI Act. We introduce the concept of influence, grounded in philosophical and chronological analysis, to examine the underlying relationship between the theories and the Act. As a proxy measure of this influence, we propose using STS to quantify the degree of alignment between the theories (influencers) and the Act (influencee). To capture intentional and operational ethical consistency, the Act was divided into two parts: the preamble and the statutory provisions. The textual descriptions of the theories were manually preprocessed to reduce semantic overlap and ensure a distinct representation of each theory. A heterogeneous embedding-level ensemble approach was employed, using five modified Bidirectional Encoder Representations from Transformers (BERT) models built on the Transformer architecture to compute STS scores. These scores reflect the semantic alignment between various theories of ethics and the two components of the EU AI Act. The resulting similarity scores were evaluated using voting and averaging, with findings indicating that deontological ethics has the most significant overall influence.

翻译：本研究通过语义文本相似性分析，探讨《欧盟人工智能法案》的伦理基础，检验规范性伦理理论与法规文本之间的契合度。尽管该法案被视为人工智能系统监管的重要进展，并强调基本权利保护，但其伦理基础仍面临道德层面的质疑。本研究考察了三种主要规范性伦理理论——美德伦理学、义务论伦理学与后果主义——对该法案的影响。我们基于哲学与历时分析提出“影响”这一概念，用以探究理论文本与法案条文的内在关联。作为影响的代理度量，我们采用语义文本相似性技术量化理论（影响源）与法案（受影响体）之间的对齐程度。为捕捉伦理一致性的意图层面与操作层面，法案文本被划分为序言部分与法律条款部分。通过对伦理理论的文本描述进行人工预处理，减少语义重叠并确保各理论的独立表征。研究采用异构嵌入层集成方法，基于Transformer架构构建了五个改进型双向编码器表征模型，用以计算语义文本相似性分数。这些分数反映了不同伦理理论与《欧盟人工智能法案》两个组成部分之间的语义对齐程度。最终通过投票与平均机制对相似性分数进行评估，结果表明义务论伦理学对该法案具有最显著的整体影响。