To make Artificial Intelligence (AI) systems and services accountable and regulated in the European Union market, in April 2021, the European Union Parliament published a proposal `Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act)', widely known as Artificial Intelligence Act (AI Act). Since then, many concerns have been raised in terms of compliance and whether the regulations are enforceable. However, to the best of our knowledge, none of them provided an explicit technical analysis of the challenges in enforcing the regulation. Among 85 Articles in the AI Act, we emphasize on the Article 10, the central regulatory requirement for data and data governance. In this paper, we have analyzed a specific requirement, the data quantity, to show the challenges of enforcing this requirement in a principled way. In our analysis, we have used deep learning modeling and machine learning generalization theory.
|Title of host publication||Proceedings of the 2022 1st International Workshop on Imagining the AI Landscape After the AI Act (In conjunction with the first International Conference on Hybrid Human-Artificial Intelligence)|
|Number of pages||10|
|Publication status||Accepted/In press - 1 May 2022|
- Trustworthy Artificial Intelligence