By Luca Bertuzzi | EURACTIV.com
08-11-2022 (updated: 10-11-2022 )
The definition of Artificial Intelligence is one of the most controversial aspects of the EU’s AI Act. [Feng Yu/Shutterstock]
Languages: Français
The definition of Artificial Intelligence and how the new EU’s rulebook for this emerging technology will be implemented will be the focus of a political meeting on Wednesday (9 November), according to a draft agenda seen by EURACTIV.
The AI Act is a flagship legislation to set the rules on Artificial Intelligence proportionate to their potential to harm people or properties. The discussions on the upcoming regulation have mainly progressed on the technical side in the past weeks.
However, more sensitive political topics need to be scaled up to the political level for political groups to reach a common position. The previous political debate occurred in October over the thorny facial recognition issues and the regulation’s scope.
EU lawmakers held their first political debate on the AI Act on Wednesday (5 October) as the discussion moved to more sensitive topics like the highly debated issue of biometric recognition.
AI definition
On the agenda for Wednesday’s discussion is the definition of AI systems, a fundamental building block to making the regulation future-proof since this technology is still at a relatively early stage of development.
In this regard, the most significant change is that the definition was moved from Annex I to an article. In the original proposal, the European Commission was empowered to modify the annexe later, an option that would no longer be available in this new setting.
EURACTIV understands that this change was because the definition is considered a part too crucial of the regulation to be changed via secondary legislation. The co-rapporteurs seem determined to ensure the definition is made future-proof via an outcome-oriented approach focused on what AI does rather than what it is and its techniques.
The rewritten article introduces three cumulative conditions. To qualify as AI, the system should be able to: receive machine or human-based data; infer how to achieve a given set of goals using learning, reasoning or modelling; generate outputs in the form of content, predictions, recommendations or decisions influencing the real or virtual environment it interacts with.
A final specification adds that “AI systems can be designed to operate with varying levels of autonomy.”
The AI definition will likely need further political discussions, as the European Parliament is still far from a shared view. Conservative MEPs, in particular, are pushing for a narrower definition of AI.
What might be able to progress faster is the second point on the agenda: the governance structure.
Each political group of the European Parliament submitted a few hundred amendments to the upcoming AI Act, setting the tone for future discussions.
AI Office
As EURACTIV reported last Friday (4 November), the co-rapporteurs proposed creating an AI Office that would play a centralising role in the enforcement architecture, notably as the body would have the power to issue binding decisions in competencies disputes between national competent authorities.
The arrangement is based mainly on the European Data Protection Board, the body that gathers data protection authorities to coordinate the enforcement of the General Data Protection Regulation. At the same time, the idea is to address what is deemed as significant flaws in the GDPR’s enforcement.
Therefore, MEPs will be called to discuss whether the AI Office should be independent and have a legal personality, its own funding and be adequately staffed, a robust advisory forum with ample stakeholders involvement and the capacity to decide on supervision disputes.
The lawmakers spearheading the work on the AI Act launched the idea of an AI Office to streamline enforcement and solve competency disputes on cross-border cases.
Political pressure
While several fundamental questions are a moving target in the European Parliament, the EU Council is about to formalise its position on the AI Act in a Committee of Permanent Representatives meeting on 18 November.
According to a European Parliament official, the fact that the other co-legislator has finalised its position will significantly increase the political pressure on EU lawmakers to reach an agreement. By contrast, a second official dismissed the allegation that political pressure was mounting.
“The co-rapporteurs know that with the current Council position, it’s important to have a solid majority behind you and properly discuss the proposals to make sure you have that solid majority,” a third parliamentary official said.
[Edited by Alice Taylor]
Languages: Français
Leave a Reply
You must be logged in to post a comment.