No longer having to present the QR code of the #TousAntiCovid application in everyday life has this joyful je-ne-sais-quoi that would push you to radically forget it. Error. On the contrary, it is time to finally understand the protocol, thanks to the eight pages devoted to it in Medicine and artificial intelligence. A book of public safety, with 33 articles by 70 contributors, published by CNRS Editions. Inria researcher Vincent Roca explains in an educational way this so-called “centralized” Covid-19 tracking technology that France has adopted, different from that “decentralized” of the British.
Whether we trust AI or reject it, because giving the feeling “of being dispossessed of control of events”, as sociologists Gérald Bronner and Laurent Cordonier write, we must read all or part of these 350 pages, concerning one of the major fields of application of AI. They were born from the success of Health and artificial intelligence, published in 2018, which prompted its co-editors, the mathematician and deputy Cédric Villani and the surgical oncologist Bernard Nordlinger, this time with Olivier de Fresnoye, at the head of echOpen factory, to put the work back on the job. Well took them. This book is concrete.
Multiple ethical issues
Doctors, hospital directors, robotics and big data specialists, psychiatrists… in fields as varied as oncology, ophthalmology or dermatology explain without pathos how the machine helps them in their analyzes and decision-making. Not naive, either. The last part of the book, which includes four of them – professions, tools, education, society – insists on the need to make the systems more transparent, to take into account multiple ethical questions, to rebalance the ratio of women to men among professionals of the field – 22% for the first, 78% for the second in the world, recalls Laurence Devillers of the CNRS, professor of AI at Sorbonne University… Trust is at this price.
Read alsoThe Challenges selection of 100 start-ups to invest in 2022
Especially since we touch on the painful. As the articles relating to oncology remind us, radiomics, that is to say the automatic processing of medical images, has taken a major place. But beware of simple beliefs.
It is not enough to have images and computers, insists oncologist Alain Livartowski, in charge of e-health programs in the data department of the Institut Curie. Many essential steps (selection of representative images, annotation, etc.) remain, which “still rely heavily on human expertise”, he writes. With the proliferation of such remarks, the reader grasps the balance to be respected between what the machine knows how to do better than us and what we humans, with our virtuoso brains, know how to do better than it. And the book acts as an anti-anxiety medicine.
(Eric Tschaen/Rea)
We wish to give thanks to the author of this write-up for this awesome web content
