- Open administration
The Open Administration Awards 2024 will be presented on May 20 in El Prat de Llobregat: save the date
The European Commission has published some detailed guidelines on the definition of artificial intelligence (AI) systems to facilitate the application of the Artificial Intelligence Regulation (AI Act). These guidelines are intended to assist providers and other interested parties in determining whether a software system constitutes an AI system as defined in the AI Act.
Key elements of the definition of an Artificial Intelligence system
- Machine-based system: AI systems are developed and run on machines, including hardware and software components. This includes everything from traditional computing systems to emerging systems such as quantum computing.
- Autonomy: AI systems are designed to operate with varying levels of autonomy, that is, with some degree of independence of action from human intervention. This excludes systems that operate exclusively with manual control.
- Adaptability: AI systems can exhibit self-learning capabilities after deployment, allowing the system's behavior to change while in use. This adaptability is not a prerequisite for all AI systems.
- AI system goals: AI systems are designed to operate according to one or more goals, which can be explicit or implicit. Explicit goals are directly coded by developers, while implicit goals can be inferred from the behavior of the system.
- Inference to generate results: An AI system must be able to infer, from input data, how to generate results such as predictions, content, recommendations, or decisions. This inference capability is a key characteristic that distinguishes AI systems from other types of systems.
- Results that can influence physical or virtual environments: AI systems generate results that can influence both physical and virtual environments. This includes everything from predictions and recommendations to content generation and automated decisions.
The definition of an AI system under the AI Act is broad and flexible to accommodate the rapid technological developments in this field. Not all software systems are considered AI systems; only those that meet the seven key elements outlined above are. These guidelines are not binding, but provide important guidance for the interpretation and application of the AI Act.