Lets have a look at the definition of „AI-System“ under the EU AI Act today. Why is this relevant? Because the AI Act applies only if an organization develops or uses an “AI-System”. 

Art. 3 (1) defines: ‘AI system‘ is a machine-based system designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.

The EU wants their definition of “artificial intelligence” to be future- orientated, and therefore must cover a wide range of data analysis techniques. That means the EU will consider not just deep learning and complex applications such as self-driving cars or systems like ChatGPT as AI. The proposed definition is so broad that many of the technologies used a variety of businesses today will fall under its regulations.

Algorithms are automated instructions and can be simple or complex depending on the number of layers the original algorithm involves. It’s a calculation process according to a specific repeating scheme. This is not (yet) an AI system. “Infers”, “autonomy” and “self learning” are the crucial words in the definition - the capability to infer and to derive models or algorithms from inputs or data which include machine learning and operated with a certain autonomy without human intervention. AI Systems are sets of algorithms that differ based on whether the data they receive is structured or unstructured. In contrast to the Algorithm they are able to deal with unexpected happenings. AI can deal with unstructured and unknown data. 

Example: A chatbot that that does not learn and only chats following a clear transcript/decision tree is no AI-system. If the chatbot learns, develops and handles questions that are not in a transcript, such chatbot will be considered AI, even if the boundaries (e.g. handover to human) of the bot are set.