AI and ethics: the heartbreaking misunderstanding of Cédric Villani
So with the Villani report on artificial intelligence, France has returned to an old tradition: asking someone intelligent to write a silly report. Anyway, stupid, we’ll understand each other: the report that our Fields Medal has just drawn up is not so stupid as agreed. We are behind on AI, quickly a national plan. Subsidies, initiatives, an agency, plenty of petits fours and picnics, the French routine what. The mountain gave birth to a mouse, the Chinese are having a good time. But the weaknesses of this report have been skillfully pointed out by others, no need to come back. What seems important to me, however, is the link that the report makes with ethics.
The very title of the report “Making sense of AI” is problematic. When we look at the history of innovation, the meaning has always been given a posteriori. There is a very simple reason for this: technological breakthroughs always present unprecedented legal, social and ethical situations. It is very difficult, if not impossible, to think through these ruptures before they occur, and before the effects are visible. We risk thinking in a vacuum. When McKinsey conducted market research for AT&T in 1989 to assess the potential of mobile telephony, the results were disastrous: no one saw the point of having a mobile phone. No one can just imagine what we would do with it. Only actual use has revealed the possibilities of the technology, just as today only the use of Facebook reveals the dangers to privacy.
More generally, the applications of a new technology are impossible to anticipate. When French and Austrian engineers discovered ultrasound in 1911, they used it to detect submarines. Forty years later, this technology is used in medicine, it is ultrasound. This use is completely unforeseen and besides, it was initially question that it is for the detection of cancers. Today, ultrasound has become commonplace and inexpensive, so much so that it is used in poor countries, especially China and India. Used for selective abortion, it is directly responsible for the fact that about 25 million women were not born in China, causing an imbalance of the sexes which leads to serious social and therefore political problems. Who would have thought that a technology developed in Europe for anti-submarine warfare was the cause, a century later, of social upheaval in Asia? Think about the consequences of ultrasound a priori would have been totally futile.
But there is worse. All technology is dual, in the sense that it can be used to do good as well as evil. Imagine that you are Minister of the Environment in an ethical country which has put the precautionary principle in its constitution (fictitious example of course). A group of industrialists comes to see you to obtain the prior authorization necessary for the marketing of their new technology. It will obviously bring immense benefits, making life easier for many inhabitants. Its only flaw: it will kill around a million people a year around the world. What are you doing? You will probably ban it and put an ethics committee on the file. This technology? It’s the automobile.
By placing AI at the service of ethics, the report therefore commits two errors: on the one hand, it does not give itself any chance to think the ethics of AI correctly, because we will think in a vacuum – we cannot think that by doing, and on the other hand he condemns France to watch others dance from the balcony. Antoine Petit, CNRS boss at the AI For Humanity conference where the Villani report was presented, invited us to avoid a pitfall: “Do not become specialists in ethics while the Chinese and Americans are becoming specialists in ethics. business. “That’s the whole point and to want to put AI from the outset at the service of diversity, gender equality, the common good and public services, is to sacrifice to the fashions of the moment with the wrong fight. We asked Cédric Villani to tell us how France could catch up in AI, that is to say to pose an industrial reasoning, not to signal its virtue to the post-modernist intelligentsia which governs the thought of this country.
Not to mention that, as is often the case in such cases, the meaning given to ethics is very limited. It may be ethical not to want to develop AI with negative consequences, but it can also be ethical to try to see, because only by doing something we will know. Entrepreneurs have known this for a long time, our intelligent scientists and those who govern us ignore it, and gradually condemn themselves to paralysis through excess of caution and, deep down, fear of the future. We are becoming an old country, and gradually let others develop the future. Basically, the Villani report is an old man’s report, the hype of our friend Cédric on top of that.
For a good critical analysis of the Villani report, see Olivier Ezratty’s article here: What the Villani Report reveals.
Philippe Silberzahn is professor of entrepreneurship, strategy and innovation at EMLYON Business School and associate researcher at thePolytechnic School (CRG), where he received his doctorate. His work focuses on how organizations manage situations of radical uncertainty and complexity, from an entrepreneurial perspective with the study of the creation of new markets and new products, and from a managerial perspective with the study management of disruptions, strategic surprises (black swans) and complex problems (“wicked problems”) by large organizations.