This article presents a conceptual framework for "normware," a proposed third level of computation beyond hardware and software, aimed at addressing the limitations of code-driven and data-driven law in regulating artificial devices within human institutions.
The author argues that both code-driven and data-driven law, while seemingly offering legal certainty, are susceptible to challenges. Text-guided law, despite its reliance on open-textured concepts and multi-interpretability, allows for contestability and adaptation, a feature missing in purely computational approaches.
The article proposes "normware" as a solution, encompassing computational artifacts that regulate behaviors, qualifications, and expectations. Normware artifacts can range from access control policies to machine learning models, all interpreted as coordinating mechanisms within a larger socio-technical system.
The author emphasizes the importance of viewing normware as a process, acknowledging the dynamic and often conflicting nature of directives within complex systems. Drawing parallels with second-order cybernetics, the article highlights the need for feedback loops and conflict resolution mechanisms to manage competing directives and ensure system viability.
The article concludes by outlining research directions for normware, including the development of languages for specifying normware artifacts and mechanisms for resolving conflicts between them. The author stresses the importance of a collaborative approach, involving stakeholders from various disciplines, to ensure the responsible and effective integration of artificial devices within human institutions.
На другой язык
из исходного контента
arxiv.org
Дополнительные вопросы