The machines are coming and only the European Union can save us
You are right, this title might be overly dramatic and paints a picture we all know from action movies. Artificial intelligence (AI) is gaining increasing prevalence in our daily lives: from surveillance to shopping, manufacturing to medicine, and producing to agriculture – AI is utilized essentially everywhere. AI is no longer a notion of what the future holds – it has become reality. However, AI remains a confusing – and often feared – topic. Citizens’ trust in AI systems is low, and the EU itself has expressed some of the fears linked to AI and its hypothetical attributes. However, with the recent publications of the AI Act, the European Health Data Space and the Artificial Intelligence in a Digital Age report, the EU is leaving its reactionary path behind and becoming proactive by building trust and ensuring that the digital transformation is human-centric and consistent with fundamental human rights.
The European Union has started to recognize the potential of AI to find solutions for many of society’s problems. In April 2021 the European Commission published its proposal for an Artificial Intelligence Act (AIA). The Union employed its most powerful legal tool by proposing it as a regulation binding in its entirety. This regulation intends to ensure that any AI technology is of high quality while safeguarding the safety and fundamental rights of citizens and businesses. Therefore, a risk categorisation is introduced which orders AI according to its potential risk. High-risk AI technology is subject to conformity assessment and needs to comply with AI requirements. The ultimate goal is to improve human well-being through AI and increase trust in the technology. The European Parliament released its draft proposal for the AIA in April 2022 and is now discussing amendments.
The AIA follows the Parliament’s previous work via the special committee on Artificial Intelligence in a Digital Age (AIDA), which completed its mandate this May when the Parliament adopted its own-initiative-report on AI. The report presents the Parliament’s perspective and highlights the essential points necessary for the EU to become a global leader in digital technologies. It underlines how the fears linked to AI outperforming human intelligence are not realistic and notes that the potential risks of AI refer much more to the introduction of biases in, for example, racial and sex discrimination in algorithms without appropriate safeguards in place. Importantly for health, the report highlights six case studies in which AI could – and does - provide significant benefit, such as health and climate change.
The brand newly released proposal for a regulation on the European Health Data Space builds on the digital health momentum and takes the use of health data to the next level. It clearly sets the future public health agenda and highlights the re-use of health data for the purpose of training AI technology. It stipulates how the secondary use of health data in training AI algorithms is fostered under the condition that it contributes to the protection of health or care for citizens. The AIA in combination with the European Health Data Space regulation could immensely boost innovation in the area of healthcare, which consisted in the past out of fragmented data collection systems and the inability to process them.
While the future of AI in the area of health seems brighter than ever, experts raise several concerns when it comes to the training of those systems and the protection of individual rights. The field of medicine and healthcare is special as it concerns our most vulnerable individuals and their data. For example, the introduction of racial biases through the training of AI systems in every field of application is stressed as a major concern. If AI systems are trained with biased data, then their results can only be the same. The solution to this is however not as obvious as one might assume. The EU has shown to be aware of this hazard and puts AI systems susceptible to biases under the high-risk category. However, the problem seems to be more complex than this. It is stipulated how medical devices shall fall under the high-risk category. Health applications and fitness apps are not considered to be medical devices and are consequently not subject to conformity assessments. Data protection experts fear that the AI Act does not adequately address the protection of individual rights and rather focuses on companies. Additionally, it is criticised how the AI Act misses the personal rights of citizens as it is stipulated in the GDPR.
The EU believes AI technology will induce the fourth industrial revolution and wants to be at the forefront of this development. The shift from a reactive approach towards a proactive one in shaping and regulating digital technologies is a step in the right direction. Nevertheless, the past has shown how the EU moves slower than its technologies. The Parliament has already warned that the EU is moving too slow, and we are being outrun by the USA and China when it comes to investments and innovations in AI. While we compete in this race for digital leadership in AI, we cannot lose focus on what really matters – harnessing the full potential of AI for the goal of improving well-being and protecting fundamental rights. The negotiation of the AI Act, which will start in the coming months, will show the direction the Union is taking and whether we are able to increase the trust in –and how much we benefit from - AI.
Happy (?) End