After the European Parliament’s adoption of the Final Report on Artificial Intelligence in a Digital Age, the European Union might position itself as the first world power to have its own legislation on AI that would apply uniformly in all Member States of the European Union.
Indeed, the EU is aware of the relevance of AI in the new technological era. For this reason, the Commission and the European Parliament want to boost the AI-based industry by means of regulation, including the risks and challenges associated with it.
Accordingly, the AIDA Committee started working in September 2020 on the impact of AI on the EU economy and its different sectors, analyse third countries' approach to AI and chart the direction for the EU. The results were incorporated into a final Report that aims to establish an AI roadmap until 2030.
The Report, adopted with a majority of 495 votes in favour, indicates the parameters of AI in the industrial sector, particularly on that the use of AI should focus on the enormous potential of technology to complement human labour, and the need for regularisation in order to avoid potential risks against human rights.
The same Report also points out that the EU has lagged behind, developing and investing much less than leading economies such as the US or China in the digital market. All this may lead to the risk that European stakeholders will be marginalised in the development of global standards and technological advances.
To this end, the Report concludes that AI, combined with the necessary supporting infrastructure, education and training, can increase capital and labour productivity, innovation, sustainable growth, job creation and thereby positioning the EU as a powerful competitor in the fourth industrial revolution.
The AIDA Committee and the Parliament are confident that the growth of the AI sector is supported by a consistent regulation of AI risks. Thus, while it is true that most AI systems in use today are low risk, certain scenarios of use may be classified as risky. Such scenarios require effective regulatory action and safeguards.
In this regard, the AI Act would not differentiate AIs on the basis of their sector, or the technology used, but on the basis of the risk that may affect fundamental rights, such as non-discrimination or privacy. Accordingly, in the case of AI systems involving high risks to fundamental and human rights, full human oversight and regulatory intervention would be required.
Indeed, the AI legislation would have specific mechanisms for human oversight in cases of risk. The Report itself discusses in detail six AI scenarios representing some of the most important AI cases in the sector, namely health, the green deal, foreign policy and security, competitiveness, the future of democracy, and the labour market.
By way of example, among the risks in the healthcare sector, it highlights that automated decision-making in healthcare applications may pose risks to the welfare and fundamental rights of patients and stresses that AI should therefore have a supporting role in healthcare, where professional human supervision should always be maintained. It stresses notwithstanding the potential of AI systems to alleviate the burden on healthcare systems and healthcare professionals, in particular, the use of safe and efficient AI applications for administrative tasks that do not require human action will save a significant time for healthcare professionals.
All in all, the final legal text still has a long journey before it is ready to be adopted. At present, we are still in a first phase, studying the concepts that should be outstanding.
The report will contribute to the upcoming parliamentary work on AI, in particular the AI Act, which is currently under discussion in the Committee on the Internal Market and Consumer Protection and the Committee on Civil Liberties, Justice and Home Affairs.
The draft AI Act will return to European Parliament at the end of this year and should be finally adopted in early 2023. It is then that the European Union will be able to claim the world' s first legislation on Artificial Intelligence.