April 2022: European Countries Unite to Support Responsible AI Governance
Summary
According to a recent report by Reuters, France, Germany, and Italy have collaborated to produce a joint paper aimed at promoting responsible governance of Artificial Intelligence (AI). This proactive approach is expected to expedite negotiations on this matter at the […]
According to a recent report by Reuters, France, Germany, and Italy have collaborated to produce a joint paper aimed at promoting responsible governance of Artificial Intelligence (AI). This proactive approach is expected to expedite negotiations on this matter at the European level.
While specific details of the paper have not yet been disclosed, the governments of France, Germany, and Italy have expressed their support for „mandatory self-regulation through codes of conduct“ for AI foundation models. However, they have also voiced their reservations regarding the use of „un-tested norms“.
The joint paper emphasizes that the AI Act primarily regulates the application of AI, rather than the technology itself. It highlights that potential risks associated with AI systems lie in their application rather than the inherent technology.
Furthermore, the paper proposes that developers of AI foundation models should define model cards, which provide comprehensive details about the construction and functioning of the machine learning model. These model cards will be based on best practices within the developers’ community and will help stakeholders understand the capabilities and limitations of the model.
To ensure compliance, the paper suggests the establishment of an AI governance body that can develop guidelines and oversee the application of model cards. While the joint paper does not currently outline any sanctions, it does open up the possibility of introducing a system of sanctions to address any violations of the agreed-upon code of conduct.
German officials have expressed their satisfaction with this collaboration. Volker Wissing, Germany’s Digital Affairs Minister, believes that this agreement allows for regulation of AI applications while preserving the innovation potential of the technology. Similarly, Franziska Brantner, Germany’s State Secretary for Economic Affairs, describes the joint proposal as a balanced approach that promotes both technological advancement and legal clarity.
This development follows the European Parliament’s approval of the AI Act in June 2021, making the region one of the pioneers in implementing legislation specifically tailored to govern AI at this level.
FAQ:
Q: What do France, Germany, and Italy support regarding AI foundation models?
A: France, Germany, and Italy support „mandatory self-regulation through codes of conduct“ for AI foundation models.
Q: What is the focus of the joint paper?
A: The joint paper emphasizes that the AI Act primarily regulates the application of AI rather than the technology itself.
Q: What are model cards?
A: Model cards are comprehensive documents prepared by developers of AI foundation models to provide details about the construction, functioning, capabilities, and limitations of the machine learning model.
Q: Is there a plan for sanctions in case of code of conduct violations?
A: While the joint paper currently does not include any sanctions, it opens up the possibility of introducing a system of sanctions to address violations of the agreed-upon code of conduct.