The use of AI technologies in healthcare needs regulation
#health #medtech #security #software #ai #neuralnetworks #mhmconsulting.fr healthcare #mittechnologyreview
While artificial intelligence (AI) is increasingly present in the field of health with the robotization of tasks, the analysis of complex problems and the support for decision-making, the discussions revolve around the use of data: how it is transmitted and how it is processed. Indeed, any digital system using health data must meet security, confidentiality, reliability and transparency requirements. However, for systems using AI, it is likely to increase bias when it is misapplied and unregulated. On the other hand, machine learning computer systems using neural networks are sometimes very complex to understand due to their continuous evolution, and are also totally opaque in their mode of operation. These different aspects translate into distrust of AI and therefore explain the reluctance to adopt it. It is therefore important to move towards a regulated trusted AI.
It is within this framework that the European Union (EU) is planning new legislation aimed at curbing the worst harms associated with IA. Melissa Heikkilä (senior reporter at MIT Technology Review), in her article published in MIT Technology Review, welcomes this initiative which responds to economic, technological, political and ethical challenges. It also puts into perspective the European approach in the context of the international AI market, dominated by the USA and China, extremely productive and unregulated. This work should even inspire the Biden's administration which seeks to bring more regulation in this area.
It will be fundamental to have convergence in the regulation of AI at the international level, because a non-alignment would pose the following problems:
• More abundant and stricter regulation may put small European MedTech companies at a disadvantage against their very rich and active American and Chinese competitors.
• Regulations considered too restrictive can be a barrier to entry for non-EU software and application manufacturers using AI, thus limiting the dissemination of these technologies in the EU.
• Additional complexity and vigilance in the operation of interoperable systems and infrastructures using AI whose locations are in the EU and outside the EU.
Link to the article