Quelle éthique pour l’intelligence artificielle en santé ? EN
What ethics for artificial intelligence in health? ALIA SANTé Integrating artificial intelligence into the medical field means asking the right questions, especially at the ethical level. Reading time: 3 minutes How to guarantee ethics for artificial intelligence related to health? Ethics is the big question related to artificial intelligence, and especially in the field of health. Indeed, what is the legitimacy of the algorithm in medical decisions with all that they imply? Should we really allow artificial intelligence to have all the pragmatism it is capable of? The question arises in particular in decision-making for heavy operations. The main question in medicine, and therefore applied to AI in the medical field, is the risk-benefit balance. We have been dealing with the risk-benefit balance a lot lately. It’s about taking into account all the factors that go into a medical choice. Thus, we get as close as possible to the pragmatism necessary to make an important decision. An understandable and explainable AI. We must ensure, from the outset, that artificial intelligence makes the right choices. Thus, the data scientist must direct it in moral choices and decisions. This is also done in the field of ethics related to artificial intelligences. Indeed, the question begins at the very moment when AI will intervene in human-related decisions. Medical coverage, hiring or access to credit. In the field of health, it is essential that the AI used is understandable and explainable. Indeed, when it comes to health and therefore sensitive decisions, it is difficult to accept the use of technologies that we do not understand. Putting artificial intelligence in place in the field of health is to ensure reliable assistance to medicine. Today, AI is not intended to replace health professionals, it aims to assist them. By inserting the IAM in the health system, it is also a daily comfort that is assured to professionals. Therefore, it is necessary that healthcare professionals intervene in the development of AI algorithms. Similarly, they will have control over all decisions related to human lives. This is how the balance between AI and caregiver, but also technology and morality will be maintained. Protection of privacy and medical data. In medicine, one of the main rules of ethics is to treat anyone who requires medical intervention. It is on this principle that medical artificial intelligence must be based. Protecting data and preventing all types of stereotypical bias is crucial to ensuring IAM performance. There’s a lot of sensitive data in the health world. That privacy data requires a lot of protection. This is why the CNIL oversees the advances made around digital health. To this end, the GDPR (General Data Protection Regulation) came into force on 25 May 2018. The purpose of this regulation is to protect users and regulate the use of data. Thus, it guarantees an ethical use of artificial intelligence in health. Protecting personal data, supporting innovation, preserving individual freedoms” www.cnil.fr Ethics and artificial intelligence, inseparable notions. In Europe, the parliament is the first in the world to tackle ethical legislation in artificial intelligence. If one thing was highlighted during the talks, it is the prioritization of human rights in the race for progress. In order to guarantee an ethical development of digital technology, the European Union, concerned about many other aspects of the web, is working to establish means of trust in order to develop rules governing the use of artificial intelligence. Today, there are very explicit sites and very relevant moral dilemmas about the choices that an artificial intelligence must make. Starting from the idea of the automatic car, the site Moral Machine has fun offering humans some of the millions of choices that an AI can make today. Thus, from the development of the idea of artificial intelligence until today, all areas are actively applied in research related to this new revolutionary pillar.
Le numérique responsable EN
Responsible digital technology ALIA SANTé Computers and digital technology play a key role in our society, and are used in almost every sector, including AI. Although digital technology has many advantages, it also causes serious damage: environmental pollution and social problems. Is a responsible digital future possible? Reading time: 4 minutes How can we preserve the use of digital technology while limiting its pollution? Can artificial intelligence using computers and digital technology be compatible with responsible digital technology? Great! A new smartphone has just been released, with new operating software, lots of different applications, frequent updates to stay connected, impressive screen quality and 5G to enjoy the Internet everywhere and without limits, to stream videos for example! I’m buying! Digital technology is constantly evolving, to the point where it is omnipresent in our society, with ever more sophisticated digital terminals and ever more functionalities. The digital industry essentially encompasses digital terminals, data centers and network infrastructures. Today, everything is changing and must move fast. With this in mind, there’s only one watchword: innovation. However, this high demand for digital innovation is not without consequences. Environmental impact Omnipresent, yes, even in environmental disasters. Energy consumption and CO2 emissions are exploding: the digital sector accounts for 4% of global greenhouse gas emissions, more than the aviation sector. Moreover, stocks of metals and ores are depleted, and impressive quantities of water are used every day: 1500 L of water are needed to design a single computer. And yet, 45% of the functions required on digital terminals are never used, and the lifespan of products is deliberately reduced in order to increase their replacement frequency. (Digital obsolescence). And what about artificial intelligence? As you’d expect: the use of large volumes of computation to learn algorithms, voracious use of servers, storage of large amounts of collected data… Artificial intelligence, while driving positive developments in the medical field in particular, is also contributing to the depletion of environmental resources and the development of greenhouse gases. Digital pollution, i.e. the pollution generated by greenhouse gas emissions, chemical components and electronic waste, is therefore increasing all the time. Studies show that the majority of digital pollution occurs during the manufacturing process. As well as a definite environmental disaster, there are also human tragedies associated with the manufacture of terminals: toxic emissions, illegal financing, pollution of drinking water sources and soil impacting food production… Environmental impact Digital technology is not available to everyone. More and more, the Internet is becoming an indispensable part of our society: administrative procedures, online payments, dematerialized information… However, the use of new technologies goes beyond and/or is inaccessible for some people, mainly in the following situations. Personal situation (age, education…)Financial situationDisability, visual impairment… To be excluded from the digital world is to be socially isolated. Although these situations are improving, they are still all too common. Clearly, there is an urgent need to implement appropriate solutions to reduce the environmental and social impact of digital technology. Today, digital technology is still too irresponsible. One solution : responsible digital So what is responsible digital? Responsible digital designates digital technology that respects the environment and its users. In other words, a digital world that is sober, sustainable, ecological and inclusive of all populations, while genuinely serving them. On the one hand, Responsible Digital encompasses all approaches aimed at improving the environmental and social footprint of digital technology (Green IT), and on the other hand, improving the environmental and social footprint through digital technology (IT for green). Le Green IT désigne l’ensemble des technologies et outils qui permettent aux entreprises de diminuer l’impact environnemental de leur industrie numérique. Le It for Green désigne les démarches mises en place grâce au numérique dans le but de sensibiliser et réduire l’empreinte écologique d’un appareil. Green AI and AI for Green Artificial intelligence, as a major energy consumer and CO2 emitter, must crucially become responsible. Green AI is possible The trend is towards high-performance models such as Deep Learning, whose applications are greatly improving everyday life, with image and voice recognition in particular. However, the more elaborate the models, the more input data, computing power and training they require. Switching to green artificial intelligence means raising the awareness of all those involved in AI design, and opting for the best possible alternatives to reduce environmental impact. Optimizing AI execution frequencyOptimization of machine learning tasksCode optimizationChoosing less energy-intensive data centersFavoring Edge ComputingImproved data management: define why we store data, and for how long? Is it really useful? AI for green Green AI is undeniably possible, and in line with the IT for Green approach, it can also help reduce environmental problems. AI, a solution to the climate crisis Par exemple, certaines IA peuvent être entrainées afin de prédire les émissions de CO2 et la consommation d’électricité, permettant de réduire l’empreinte carbone. Dans cette continuité, l’IA peut aussi être utilisée pour mesurer l’élévation du niveau de la mer, détecter la présence de plastique dans les océans et contribuer à la sauvegarde des coraux. So to speak, AI has a wide range of possible uses in our society, in sectors as diverse as agriculture, healthcare, geothermal energy and aviation, making them more responsible and greener. Levers and tools for digital responsibilit To support this transition to responsible digital services, a number of tools and means of action to be adopted on a daily basis are being shared and implemented. For example, the Institut du Numérique Responsable INR has published a reference guide (#GR491), divided into 8 families and offering 57 recommendations for the design of responsible digital services. Responsible Design Reference Guide (institutnr.org) GR491, The Reference Guide to Responsible Digital Service Design | INR (isit-europe.org) Reduce carbon footprint Limit, compress and adapt image and video formatsLimit the number of fonts usedLimit Mail exchanges and delete them regularlyLimit cloud computingChoose a responsible web hostOptimize and better manage our data Reducing the impact of our equipment Implement a
