On Tuesday 14 November, at the I-Com premises in Brussels, PromethEUs network presented its joint paper on the EU Artificial Intelligence Act (AI Act) as the main output of its activities in the second semester of 2023.

The paper was presented at the Conference titled “AI and the future of EU competitiveness: the road ahead”, which was an interesting opportunity to discuss the state of the art of the EU AI Act.

Aiming to contribute to the debate around the topic from a Southern European perspective, the paper set the discussion on the latest developments of the dossier ahead of the agreement that negotiators aim to secure by the end of 2023 and on the divisive subjects that remain (e.g., the use of copyrighted content by AI and biometric surveillance).

Chapter 1, beyond providing an overview of four big international initiatives and approaches to AI regulation, focuses on the EU AI Act, arguing that crucial issues still need to be agreed on, i.e., those under discussion in the trilogue and the terms and role of human oversight. The chapter points out the EU AI Regulation will need to be updated regularly in the future due to the everchanging nature and capabilities of AI, and the potential unintended legal fragmentation resulting from the AI Act subsidiarity dispositions (penalties, high-risk classification). It also lays out different areas where the trilogue negotiations will still have to establish consent between the Commission, Parliament and Council.

Chapter 2 analyses the impact of generative AI, especially from an economic point of view. The analysis describes the European competitive position (with a focus on the Southern countries of  Greece, Italy, Portugal and Spain) in the field of generative AI, and offers a global overview of the situation. Finally, it focuses on the potential challenges related to generative AI and on how to tackle them.

Chapter 3 analyses why is generative AI substantially different from AI and which countries are leading the AI race in private investments, patents, publications. The results show that the dynamics of U.S.-China competition are also being translated in the AI race, with both countries leading the rankings consistently. However, the authors show that the most-used current projects in generative AI have a global reach, as they are open-source.

Chapter 4 depicts how the usage of AI is currently low, and this presents an opportunity for a mindful deployment of such technologies to rip its benefits as much as possible while keeping under control its pitfalls. Through an insightful SWOT analysis, the authors point out that the relevant research in the network’s countries steadily increased in the last two decades while there is a significant surge of investments in AI start-ups in the last three years.

Please find here PromethEUs Joint Publication.