Data collected for creating visualizations of governance architectures established by the AI Act (regimes high-risk AI systems, general-purpose AI models and systems)
Description
The data collection was undertaken to prepare visualizations of governance architectures regarding high-risk AI systems and general-purpose AI models and systems established by the EU regulation 2024/1689 (AI Act). The data collection was part of an interdisciplinary research project including scholars from legal studies, information systems and ethics. The dataset is a collection of the following: all actor mentionings contained in the AI Act, mentionings of technological objects, systems and processes, legal relations (rights and duties) between different entities (both actors and other entity typse such as technological objects) and the governance approaches that apply to them. The data was used to create visualizations using an architectural ecosystem modelling approach. The dataset is accompagnied by a manual and codebook regarding the process of data collection and the development of visualizations.
Files
Steps to reproduce
1. Data collection and coding Legal researchers analyzed the official English version of the AI Act from the EUR-lex database, identifying relevant entities (actors, technical objects and systems) and the legal and technical relations between them. Guiding questions for the data collection: - What actors and technological systems are defined? - What legal and technical relations (rights, obligations, techical processes and dynamics) does the AI Act outline with regard to high-risk AI systems and general-purpose AI models and systems? Findings were documented in tables. 2. Entity Classification and Relations Entities were categorized into: Actors: Human individuals, organizations, institutions, bodies, or subdivisions thereof. Technological Object or System: A physical or digital artifact designed and engineered to perform a specific function or set of functions. Entity relations were classified into: Legal relations: A physical or digital artifact designed and engineered to perform a specific function or set of functions. Technical relations: Dynamics and processes between technical objects or systems. Socio-technical relation: Dynamics and processes between actors and technological objects or systems, which are not a legal relation. 3. Data analysis to prepare the visualisation The coded data was analyzed to develop a governance visualization of the AI Act. Researchers identified actors and actor categories, technological objects and systems, and regulatory provisions addressing specific AI technologies. Legal relations with regard to two governance regimes regarding different AI technologies, concerning high-risk AI systems and general-purpose AI models and systems, contained in the AI Act were identified. After examination of the identified relations, ten different types of legal relations where developed with regard to different governance approaches and each relation was classified into one of the categories.
Institutions
Categories
Funders
- Volkswagen FoundationLower Saxony, Germany