1 June 2019 - 10h42
The Institut Le Rosey had organized a conference and discussion to be held at the Yacht Club on 24th April on the subject “What role will artificial intelligence play in the world of education?” The conference, which was postponed sine die because bad weather caused some flights to be cancelled, was due to include interventions by personalities such as Luc Ferry. The philosopher, writer and former Minister of Youth, National Education and Research agreed to talk to La Gazette de Monaco remotely, however.
Our educational system essentially uses very conservative methods, but with the arrival of the new technologies, it needs to develop. Could you tell us about the revolution and methods? Will it mean a profound overhaul of teaching methods and programmes?
I’m going to disappoint you greatly, because I tend to think that the main problems in our education system derive from what was intended to be ultramodern and innovative from the 1970s due to the effect of education sciences and the heritage of the joyous month of May. I don’t believe in a methodological revolution for a second, and because, for example, the number one problem today is primary schools and the rise in illiteracy, it is by returning to fundamentals and classic learning methods that we will obtain the best results. The MOOCS (Editor’s note: online training open to everyone) have largely been a failure, and those that do work well are at a higher study level, in universities, in high schools to some extent, not at all in middle schools and even less so in primary schools, because at this level, with young children, a teacher’s real attention in the classroom is very simply irreplaceable.
What role can artificial intelligence play in this context?
Virtually none. It creates a problem for us that is anything but the teaching methods, a problem that is going to become absolutely crucial in future years: how to equip our children for the world to come, what knowledge and skills to provide them with, knowing that AI is going to impact nearly every profession in the future, including those that require higher qualifications: doctors, lawyers, accountants, notaries, etc. As far as these issues are concerned, Google’s pedagogical efforts are totally derisory. They may be of some purely technical use for facilitating communications between students and teachers, or replacing some books, but basically, they don’t achieve a great deal…
What does this phenomenon mean for teacher training?
Teachers, like politicians and parents, must understand that AI is going to bring massive changes to the labour market, and that we need to think about it now. To give you an example, driverless cars without a steering wheel are now perfectly ready from a technical standpoint. I tried one myself in Paris, and it’s impressive. This means that in 20 years there won’t be any more taxi drivers in our major cities. It has been estimated that in the next 20 years, several million jobs in the United States will be impacted by driverless trucks. But it will be the same thing, for example, for radiologists, whose profession will be very heavily affected by AI. They won’t disappear completely, but for every ten radiologists only one will be required, and the rest of the work will be done by the machine. The conclusion is that if our children are not going to be outstanding data scientists, mathematicians or biologists in the future, they would be better off going where AI won’t. The professions that use the head, the heart and the hands aren’t going to disappear. Chefs, gardeners, nurses and hotel and restaurant managers are the professions of the future, and our training needs to take account of this right now. This is what AI will change, a thousand times more than teaching methods…
What are the limitations of artificial intelligence, if any?
If we want to properly understand what is at stake here, we need to clarify the differences among the various types of artificial intelligence, and there are three of them. First, there is weak, “narrow” AI. It sequences genomes and reads x-rays better than a radiologist, it is beginning to do wonderful things in the field of automatic translation, it managers entire areas of the so-called “collaborative” economy with applications like Uber and Airbnb, an economy that is characterized above all by the fact that artificial intelligence enables non-professionals to compete with professionals in a number of sectors. This AI will have (and is already having) major effects in medicine, defence, the organization of the road and air traffic, driverless cars, monitoring dependent individuals in their homes, the fight against crime and terrorism, the organization of humanitarian aid and a thousand other sectors. To tell the truth, virtually no area of the world of labour will be spared. But it doesn’t think. Of course, it calculates brilliantly, and is even capable of ultra-high-performing algorithms, but it lacks the self-awareness that a one-year-old child has more of. The second face of AI is “super AI”, which is still a weak AI, and doesn’t have any form of awareness, but it will be able to contextualize. The peculiarity of weak AI today is that it is still virtually incapable of contextualizing the questions it is asked. It can beat the world Go champion, order an Uber, drive a car or sequence the genome of a tumour, but once it’s outside its own “lane”, it no longer knows how to do anything. Its intelligence and performance are “vertical” and only very slightly “horizontal” or transversal, which means that it sometimes has to make a considerable effort to find a solution to something a five-year-old child would be better able to arrive at. The idea that is driving AI researchers today is that it needs to be made more transversal and more contextualizing, so that it can leave its lane and become, in the words of the definition Nick Bostrom has given to “super AI”, “superior to humans, not in a single area like chess, Go or sequencing a genome, but in every area”. The third face of AI would be (and I’m using the conditional, because it seems to me to be utopian) strong AI, a form of intelligence like ours, equipped with self-awareness, free will and emotions, but incarnated (if I can use this term) in silicone rather than carbon. In essence, we might say that strong AI will be the intelligence of a machine that is capable not only of mimicking human intelligence from outside, but thanks to connections of artificial neutrons equivalent to our own, will be able to equip itself with the three elements that have been exclusively human up to now: self-awareness, the ability to take decisions and emotions (love and hate, fear, suffering and pleasure, jealousy, etc.). Strong AI is not there yet, but the two other forms of AI have clearly arrived, and these are the ones that pose the problem of the professions we need to prepare our children for.
10 July 2020 - 07h38
Support bold projects with high potential and impact. This is the objective of MonacoTech, whose last call for applications dates back to March and April, ended with the selection of…Read more
4 July 2020 - 07h41
Under the impetus of its secretary general Bernard D'Alessandri, the Yacht Club de Monaco created in September 2019 a commission dedicated to these new engines. A way of bringing together…Read more