Desafíos actuales de la Inteligencia Artificial

Analysing the interplay between data spaces and article 10 of the AI act: a case study of ... 77 develop the (to the best extent possible) debiased AI system or affects also the data space? As such, there are some potential tensions regarding the interaction of this provision with the envisaged data spaces. To answer these questions, we need to understand what data spaces are and how can they help in this challenge of addressing biases in AI systems. 3. THE CONCEPT AND ROLE OF DATA SPACES The European Commission has defined data spaces differently across policy documents and regulatory proposals (Chomczyk Penedo, 2024). In general, it can be argued that a data space is an interoperable environment where data can be shared and accessed under well-de- fined conditions, ensuring trust, security, and data sovereignty. The EU has been at the forefront of promoting data spaces through various initiatives and regulatory frameworks. The EU 2020 Data Strategy outlines the vision for a single Eu- ropean data space, where data can flow freely across sectors and borders, fostering innova- tion and economic growth. This strategy emphasizes the creation of nine sector-specific data spaces: industrial (manufacturing), Green Deal, mobility, health, finance, energy, agriculture, public administration, skills. These are pivotal in unlocking the potential of big data and AI, as they enable the pool- ing of diverse datasets, which might lead to more accurate and unbiased AI models. This can drive advancements in AI, as diverse and high-quality datasets are essential for training robust AI models (Trigo Kramcsák, 2023). In consequence, it is interesting to explore how data spaces can help in the application of Article 10(5) AI Act. While Recital 68 used as an example the European Health Data Space, there are eight other fields where the balancing needs to take place. For this purpose, we can look into the only other field where a regulatory proposal for a data space exists: financial services. 4. CASE STUDY: CREDITWORTHINESS AI SYSTEMS IN FINANCIAL SERVICES Given the wide range of applications that AI can have, as well as the varied data spaces that are intended, we will rely on a case study to highlight some of the tensions that might emerge. For this purpose, we will present the case of algorithmic credit scoring systems and their framing in the AI Act (4.1). Then, we will analyse how these systems are includ- ed in the sector specific data space, the FiDAR proposal (4.2). Finally, we will present the interplay between data spaces and detection and correction of algorithmic biases in this scenario (4.3).