On the 8th of December 2022, Dragoș Tudorache carried a compelling virtual presentation on the topic of “Governance: the Key to Success in Regulating AI,” which explored the motivations behind the European Union’s regulation of Artificial Intelligence, how it has impacted our societies and democracies and why we must address its position in our future adequately.
The discussion on the regulation of Artificial Intelligence came into focus when the President of the European Commission, Ursula Von Der Leyen, announced that the production of a horizontal regulation of AI would be prioritized during the first hundred days of the year 2019. The significance herein marks the first International Governmental Organization, which was publicly placing efforts into the regulation of AI development. It would then influence the digital transformation within societies and economies and change human-to-human, human-to-state, and state-to-state interactions, impacting democracy and the overall physical world around it.
Mr Tudorache highlights another important point: AI’s influence and regulation’s influence on geopolitical reactions. Technology is an essential driver in the global competition or, in the mind of some global actors, the race for supremacy. Moreover, he explains that this is linked to the manner in which we understand the values that underpin our society by using the case of China to demonstrate how other nations could instrumentalize technology and artificial intelligence to impose certain models of how it is run within society. For this reason, it is crucial to take a similar approach when discussing with like-minded partners to ensure that technology and its development standards remain in sync with the current shared values and democracy.
Another essential point that EU stakeholders should maintain, is being more mindful of global players and that the European Union should not solely rely on the “Brussels effect,” (which is the process of regulatory globalization triggered by the EU externalizing its laws outside of its borders due to market mechanisms), when developing an AI regulatory framework but also engage with various partners throughout the world, to provide a shared understanding and base of the rules in place, regardless of the difference of approach in legislation. This sort of AI diplomacy is paramount for success in creating a model which can be embraced and integrated by others and should not only be done on the transatlantic route but also throughout other countries and various international settings such as the OECD or G7.
Regarding the current development of an AI framework, Mr Tudorache explained that the EU Commission commenced this process with a consultation, which included all affected stakeholders before passing any legislation. This was done through the publication of a White Paper in February 2020, which opened up the discussion between various parties, which then contributed to the enrichment of the commission’s thinking, thus creating a proposal that accurately anchored the stakeholders’ expectations. As legislators, it is vital to understand the complexities of this technology in order to prepare better and negotiate the regulations.
This process was longer than previously expected and is currently behind schedule as negotiations were planned to finish by the end of 2022 due to numerous amendments by constituents, which echo society’s misconceptions, concerns or priorities. However, in March of 2023, Mr Tudorache believes that the parliament will reach a vote in the primary and begin trials as he is optimistic that both the parliament and council are in accordance with the initial proposal.
There will still be the need for national regulators due to the innovation driver effect that will be fostered through AI Sandboxes, which must be accessible to all levels of organizations and companies as possible. For this reason, the legislation itself encourages the development of AI sandboxes not only at the regional level but also at the Municipal level. This is done to achieve the fundamental core of this text, which is trustworthiness. For AI to be accepted by users and taken up by public sectors as tools of the trade, trust is mandatory and enforced through rules and standards that control its use. Despite this progress, Mr Tudorache believes that there could remain debate on the definition of prohibited practices (as outlined in the Pyramid of Risk in the Regulatory Framework) and what it entails. This is due to the provision of biometric identification in public spaces in real-time, the discussion surrounding predictive policing impacting individual rights, and finally, the lack of definition of high-risk AI technology and the general purpose of AI.
We would like to thank Dragoș Tudorache for his insightful talk and taking the time to discuss this important issue with the IEAI community.