Our blog posts will help you understand the ethical stakes and Swiss laws related to medical AI.
Since the earth-shaking created by the advent of ChatGPT four years ago, we have been hearing about artificial intelligence (AI) virtually everywhere, although, for specialists, AI had already quietly infiltrated our lives for years or decades. This recent surge of LLM has both consciously fixed AI existence and narrative in everyone’s daily life and helped realise its enormous potential. It has also sharply raised awareness about the serious risks bound to it. Now, as everyone wants to be part of the AI spree, including the biomedical world who quickly understood the tremendous potential for healthcare or discovery (and funding, since money follows innovation, regardless), the materialisation of risk becomes a global reality. In medicine, that may mean patients treated unfairly by algorithms, personal health data falling into the wrong hands, or no one being clearly responsible when an AI makes a medical error, just to list a few examples. Good news: regulations can reduce these risks and help preserve human rights, but only as long as the rules are clear, known, and people actually follow them.
And that will be the topic of a series of blog posts.
At the heart of our peregrination lies the fact that Switzerland has no specific law for AI in biomedicine, or even for AI alone anyway. In its AI roadmap, the federal Council has decided not to take the path of a potential global AI act, considering that the existing sectorial laws were sufficient, although occasionally amendable, and that sectorial recommendation would be encouraged. The recent ratification of the Council of Europe’s convention on AI by Switzerland would put regulation pressure only on public authorities, not businesses or individuals, preserving the competitiveness of the Swiss private sector. That’s a usual move in Switzerland, a blend of pragmatism, wait and see attitude, trust in collective common sense, and resistance to legislate. Therefore, the Swiss regulatory framework for biomedical AI resembles a complex constellation with interactions of multiple federal laws, and further influence of cantonal regulations, international treaties, and professional codes. The encouraging note is that they usually agree on the same logic, since they revolve around core principles such as human dignity, equality, autonomy, privacy, and accountability.
So, can we start unravelling the ball of regulations governing the development of biomedical AI systems, from their conception to their testing on human subjects? Can we shed light on the regulatory issues associated with the use of AI in medicine? Yes, and that’s exactly what we will do in the posts to come. No worries though, we will only mention the laws important in that matter, without getting too much into the nitty-gritty of their intricate application.
Stay tuned for the upcoming posts!
(Word cloud designed on https://wordclouds.ethz.ch/)