Speech by the President of the Republic Emmanuel Macron #Aiforhumanity

France Emmanuel Macron · President of France

29 March 2018 · AI for Humanity presentation · Collège de France, Paris, France

Context
Macron addressed researchers, entrepreneurs, and officials at the Collège de France on March 29, 2018, to unveil France's national artificial intelligence strategy.
Main themes
Artificial intelligence strategy, ethical responsibility of AI, technological revolution, French AI policy, Villani report, economic and social transformation, philosophical dimensions of AI, European cooperation on AI, open and collaborative innovation, Promethean ambivalence of technology
Duration
~79 min
97th percentile
Words
9,458
97th percentile
Sentences
292
95th percentile
Readability
32
20th percentile
Grade
16.7
88th percentile
Syntax
3.34
85th percentile
Vocabulary
2.13
34th percentile

Translated from French

Content
Actors
Style

The broad sphere of national life a sentence belongs to. Every sentence has an Arena, so it is the denominator the other indices are read against.

Other
  • Technological 40%39%
  • Social 15%15%
  • Institutional 15%15%
  • Economic 13%13%
  • Geopolitical 10%10%
  • Ceremonial 7%7%
  • Environmental 0%0%
  • Not in focus 0%

292 sentences in 100 paragraphs

Madam Federal Minister of Education and Research of Germany,

Mr. European Commissioner, dear Carlos,

Ladies and Gentlemen Ministers,

Mr. Deputy VILLANI,

Ladies and Gentlemen Parliamentarians,

Mr. Rector of Paris,

Mr. Administrator of the Collège de France,

Ladies and Gentlemen,

Dear Friends,

First, I want to congratulate the organizers; I think that coming to talk about intelligence, and in particular artificial intelligence, at the Collège de France is not a bad idea in itself, and it brings us back to a glorious history of our country, to profound choices that have indeed gathered around a trauma that is not appropriate to revisit today regarding the relationship between France and its university more broadly, but to gather the greatest minds in one place and seek to shed light more widely. Then, I understood, Mr. General Administrator, your message; it is appropriate that I take literally the recommendations of the distant successor of Claude BERNARD to ensure that the strategy will thus be greatly pursued; I am not far from it.

I want indeed, above all, to thank the eminent researchers who have agreed to come, some since yesterday and throughout this day, to reflect—academics, researchers, entrepreneurs, start-up founders—beyond all sectoral barriers, all the divisions that sometimes exist, all the differences in conditions to share collective reflection and, in a completely cooperative manner, help to build this strategy.

Artificial intelligence, I must say, says one who has only a philosophical training and who is therefore always very admiring of mathematicians, probabilists, and all the great scientists he has before him, refers us to what would be the Leibnizian hypothesis according to which there are several possible worlds. There is in LEIBNIZ this hypothesis that "God calculates for us the best possible world," and there is therefore something of an almost Promethean option that allows us to revisit LEIBNIZ's conception of the world through artificial intelligence, which would give us the ability to perform this calculation ourselves and, indeed, through learning machines, to traverse the paths of misfortune much more quickly to choose the right path much earlier and much faster.

It is Promethean in what it entails of ambivalence; it is an incredible opportunity to accelerate the calculation reserved for God in LEIBNIZ, and it is an enormous responsibility to have in our hands this possibility to do so. Ultimately, it is a bit of all this that we must talk about today, and I believe a bit of that which you have largely discussed. No one wants to deprive themselves of this faculty, but each considers it at the moment it is offered to them with legitimate dread, and I would say that anyone who does not feel a form of fear or does not dread this possibility could legitimately be judged unconscious.

Artificial intelligence finds its origin as early as the mid-20th century, and the explosion of computing power, the multiplication of data volumes, and the rapid rise of new algorithms have ushered artificial intelligence into a new era that is both a technological revolution but also an economic, social, and of course ethical revolution, and thus a political revolution in the fullest sense of the term, if I dare say. This revolution will not occur in 50 or 60 years; it is happening now. New opportunities are being offered to us, choices are already before us to pursue or not certain paths of innovation, to frame them, to open or not certain possibilities that are offered to us.

Therefore, we are already in this new grammar that I have just mentioned, where we must articulate the radicality of choices and deep technological, social, and economic innovation with the responsibility of ethical choices. That is why I want to particularly thank Cédric VILLANI for having produced an extremely enlightening report on this subject through the 300 hearings he briefly recalled, the preliminary reflections of France IA, which I thank, of the Parliamentary Office for the Evaluation of Scientific and Technological Choices, of the CNIL, and through this work of synthesis, hearings, and invention because reflection presupposes it, he has given artificial intelligence a precise vision, free from anxiety-inducing prejudices but responsible. In the dialogue he has been able to conduct from the beginning with the ministers, whom I thank not only for their presence but for their action from the outset of this work, he has allowed these few lines of force to emerge that I want to present before you today.

It is not a matter of having a strategy thought out from above, so to speak, coming from above that would come crashing down on what artificial intelligence is. First, it would be futile to want to do this in a single country, and secondly, it would be contrary to the very dynamic of what artificial intelligence is, one of whose strengths is precisely to have given power to the many against the one. We must therefore think about the capacity to define the rules and the strategy, which requires a moment of synthesis and democratic decision in this movement perpetually thought out by the multiplicity of actors and this permanent branching that goes through the innovation of this multiple. In other words, we must hold in one hand—at the same time, some would say—the technical progress and human progress, economic transformation and social transformation, and succeed in ensuring that this Promethean utopia does not become a dystopia.

To do this, we must first not yield to what has sometimes been a European tendency, and I would almost say often a French tendency, which is that the perfect and acute awareness of all the undesirable effects of an innovation can lead us to refuse it outright. We sometimes have this flaw; we have had it in our industrial or recent innovation life; we must not yield to it. And so I believe that the first thing we must agree on is that we must not be afraid and refuse change. First, because it is at work everywhere and because it can also lead to the best and to innovations that will allow us to live better.

Then, we must not, however, I believe the second principle we must set for ourselves, be naive and consider that change, technological innovation—which is this movement that we do not always manage to perfectly synthesize conceptually—will carry within itself the conditions for its own regulation and thus its proper articulation with the common good. That is not true, and we live in a world that is full of the excesses linked to the negative externalities of contemporary changes and to the incapacity that the multiple actors have had to think about their own regulation, to think, if I may say so, about the own inhibitor to their action sometimes that allows for the articulation of individual actions with the common good. In this matter, I do not fully believe in the perfect composition of individual interests and the invisible hand. And so we must neither be afraid nor be naive but define this ridge path that will allow us to have a strategy for artificial intelligence.

This strategy, in my eyes, involves four main elements that largely stem from the work presented by Cédric VILLANI and from what you have conceived together. The first is to put ourselves in a position to build, or rather to strengthen, in France and Europe the ecosystem of artificial intelligence and, in particular regarding talent, a true research network and experimentation. The second axis is precisely to engage in a resolute policy of data openness in order to foster the emergence in France of champions of artificial intelligence or to encourage their development.

The third axis is to have a strategy both for financing and for projects—a French and European public strategy that allows us to develop and accelerate our presence in certain sectors and to succeed fully in international competition. From health to mobility, I will return to this in a moment. And finally, it is to think about the terms of a political and ethical debate that artificial intelligence fuels everywhere in the world but whose terms we must set because we need to articulate reflection, rules, and a common understanding.

The first pillar of this strategy, therefore, is this innovation ecosystem that we absolutely need. Several major technological disruptions have indeed shaped the world in which we live today in recent years, particularly the digital disruption. Sometimes France has been slow, and in this matter—and I would say even more particularly for artificial intelligence—the delay in building or completing an adapted ecosystem is the certainty of missing opportunities and taking the risk of already losing a battle. Therefore, the capacity to improve this ecosystem means additional opportunities and more chances for our training ecosystem, research, and our economic ecosystem to succeed.

We have assets to succeed in artificial intelligence. I know how to be critical of our country when necessary and to face our own handicaps, but in this matter, we have assets because we already have talents, we have excellence in training in computer science, in mathematics—our rapporteur of the day will not say otherwise—and we have talents who have spread and who have, one and all (several are in this room), built excellence in this area in Europe and beyond Europe. These are researchers, French entrepreneurs who have contributed to the autopilot of our planes, to the most refined realizations of synthetic images that we can see, or to the most advanced theories that have enabled special effects technology. We have these talents; this ecosystem is here.

It has indeed convinced some who knew it well to return or invest in France in recent years; I am thinking particularly of FACEBOOK, GOOGLE, SAMSUNG, DEEPMIND; I thank them for their trust because they have helped to fertilize this ecosystem. It is to this excellence that we owe the announcements of establishments today; the research centers of SAMSUNG, FUJITSU, IBM, DEEPMIND joining those of GOOGLE or FACEBOOK will contribute to creating in France a network of skills, laboratories, and experiments at the highest global level, which is a tremendous asset. This asset is the French intelligence in these sectors that is responsible for it, and it is the capacity of the leaders of laboratories, schools, universities, and organizations that has allowed, over the last generations, to consolidate it, and for that, I thank you deeply.

Now, I heard the concerns expressed by some who are in this room with whom I informally discussed yesterday, which is to say we are almost reaching a point where our ability to train and retain talents in this French public ecosystem is at risk, and therefore we must hear this message and take our responsibilities so that this ecosystem can continue to develop and accelerate its growth without being weakened; that is why we need to make some decisions.

The first is that we will set up a national program for artificial intelligence. It will be coordinated by INRIA in connection with other partner research organizations and the related universities, and it will involve the entire French scientific community. We will particularly establish an emblematic network of four or five dedicated institutes anchored in university hubs scattered across French territory. Some of these hubs already exist; we are not starting from scratch. The "PRAIRIE Initiative" in Paris, in Saclay with the DATA IA institute and the DIGIHALL project, and we will develop them in partnership to have precisely this network, this very open community that will have permanent interaction and will allow for the rapid amplification and consolidation of other nodes of the future network in Toulouse, Grenoble, and other sites.

This network connected to the French university research ecosystem will be the flagship of French public research in artificial intelligence and an essential element in the training of new talents. It will aim to welcome private funding—I know how much the minister is attentive to this good partnership and to these balances—and I want to acknowledge here the industrial partners who have already wished to associate themselves with it. This will allow us to create, beyond what exists today, chairs of excellence that will also allow us to bring in for a time in their careers or more stably the best researchers or researcher-entrepreneurs in the world in this field to come and settle for a time in France. I particularly wish that these projects be directed towards our European partners and first and foremost towards Germany; that is why a portion of the funding will from the outset be earmarked for Franco-German projects.

We must also make a particular effort in training. It is indeed our responsibility to ensure that French universities provide the necessary skills in this research area to support the growth of the ecosystem. The primary challenge of artificial intelligence on a global scale will be that of talent; it already is that of talent. We have attractiveness through our excellence because we are cheaper than many others, but the risk is that there will be a rapid drying up of this talent pool that several of you have pointed out.

If our higher education system does not know how to train and attract enough technicians, engineers, and doctors to support the development of laboratories and companies in France, and to retain its talents in training new generations, we will not succeed in consolidating this conquest of a new horizon. That is why we will double the number of students trained in artificial intelligence, from undergraduate to doctoral levels, including short professional training programs, and we will provide the funding that corresponds to this doubling.

France, to maintain its rank, must also shine on the international stage. I just mentioned it; it is necessary to have individual chairs to attract the best researchers. It will also involve establishing a series of calls for projects similar to what we did with "Make our planet great again," around this initiative consolidated by the CNRS, which allowed, through calls for projects, to select the best global projects and to differentially fund the best research in this area, thus attracting the best teachers and researchers for part of their careers or for a year, attracting the best research projects through these selections to make France a place of attractiveness and international competitiveness in this field.

The excellence of our research will be the breeding ground for our future industrial champions. To achieve this, we must drastically increase the permeability between public research, which is essential, and the industrial world—perhaps even more so in the field of artificial intelligence than in others, as we have experienced during previous generations of innovations. And that is why we will carry in the upcoming PACTE law, which will be presented by the Minister of Economy and Finance, a reform, closely worked with the Minister of Research, which will eliminate the mandatory passage through the ethics committee or which will allow a public researcher to dedicate up to half of their time to a private entity, whereas today it is capped at 20%. This has led, and I have demonstrated it in recent days in discussions I have had with various individuals, far too many researchers to join Anglo-Saxon universities that allowed them this flexibility—explaining to me that it was precisely for this flexibility that they made such a choice.

It is in this cross-fertilization that we will build a true research ecosystem, and it is because we will succeed in creating this back-and-forth, because we will also create an ecosystem where entrepreneurs will see themselves as always linked to the university and the research organization—because they are somewhat the moral debtors, if I may say so, or because we allow them this flexibility—that we will be sure that our research organizations, that our training maintains this excellence and this capacity for attraction.

I also wish, and it is absolutely key for this artificial intelligence ecosystem to be fully adapted to our contemporary challenges, that we give more space to experimentation. You mentioned it, Mr. Deputy, and I say it here very clearly, we need today to improve our speed in terms of timelines and authorizations regarding experimentation if we want to fully develop our own innovations or promote them.

The time currently required to test an innovation in France, whether it is an algorithm and its final effects on a particular outlet, or a medication, is not always and entirely that of an innovation economy. I know how much the Minister of Solidarity and Health is also attached to this issue. I therefore wish that we could, in the short term, relaunch an experimentation call by this summer, in order to introduce amendments into the PACTE law that authorize the most significant experiments.

I wish that, in each sector related to artificial intelligence, if evolutions or improvements in experimentation frameworks are desired and fall under legislative provisions, we could examine them for modification this summer, as we will do in certain sectors that I will return to in a moment. Beyond that, we will fundamentally reform the France Experimentation program—Mr. Secretary of State for Digital—so that it becomes both permanent and embedded in the missions of each ministry. This ability to disseminate experimentation and innovation at the very heart of state action is an essential element for succeeding in this bet on artificial intelligence, which is first and foremost a bet on human intelligence, the ability to train, to conduct research, to move faster in it, to be able to experiment much more quickly.

It is by creating this powerful matrix that we will be able to enter the global competition where Anglo-Saxons and the Chinese have already invested massively. This lead is by no means a fatality if we manage to develop this ecosystem in France as well as in Europe, and it is this same fight for the transformation of this ecosystem that I wish to lead within the framework of the next budget, as well as in the ongoing discussions at the European level.

The second area that must be ours is that of data. You have just recalled its importance, as has the Commissioner, and it is an asset for France. We have massive centralized databases that offer us the opportunity to position ourselves at the forefront of artificial intelligence in certain sectors. Many sometimes criticize France for being a Jacobin state; this has its advantages, as it is organized, structured, and, in these moments, it can be a chance. We must therefore know how to use it.

To ensure that data is exploited to the fullest, we will methodically but resolutely proceed with a proactive opening of our data. This starts first and foremost with the opening of public data. We will continue to strengthen and expand the public data service, that is, the list of publicly available databases that are maintained with a very high level of quality, freshness of data, and availability of services.

Laws were passed in the previous five-year term that defined the framework; now we need to move on to practical work, and the Secretary of State, in the coming months, will continue, in connection with each of the ministerial departments, to open these public data and to define the explicit framework.

We will also open data funded by public funds. In particular, health data, but also data from transport operators, for research and general interest purposes, or by default, data from research projects funded by the state, for example, under the "Open Science" plan led by Minister Frédérique VIDAL. This is also a way, based on data that is consolidated by programs funded by the public, to create a framework that allows for an acceleration of our innovations and progress in artificial intelligence.

Finally, we will facilitate the creation of data-sharing platforms between public and private actors, with a sectoral logic. Indeed, economic actors themselves need to go further in their practice of sharing and valuing their data. Like networks, the value of data sets grows more than proportionally, as we know, with their size; we must therefore convince and create the framework in several sectors to open data between actors who may be in competition, but whose cooperation in terms of openness can allow for innovations, improvements in their productivity, their business models, and profoundly radical innovations, if they agree to adopt a cooperative logic based on these platforms.

In agriculture, for example, to take just one very concrete sector, we can do much better, have innovations, as we know, that are disruptive; we have started to do this in certain sub-sectors if we manage to open data between SMEs that are currently scattered, to share data, to better model and operate artificial intelligence models that will allow us to go much further and do better.

We will therefore engage in pedagogy and support all private initiatives for opening and exchanging data so that an open data economy emerges in all sectors. From the banking sector, for transaction data, to agriculture, for the efficiency of pesticides, to data from autonomous vehicle tests, I am calling for a general mobilization around data today. This work will also be accompanied by this culture of openness, the promotion of free software, wherever we can develop it.

Correspondingly, I finally wish that we could open a reflection at the European level on access, for general interest purposes, to massive private data bases, particularly those of very large actors who are in a de facto monopoly on the collection of certain categories of data. Because this is a situation rent that we have consolidated in certain sectors, creating a "squared" rent, so to speak, of data consolidation, and we need to be able to open it up, because this data has a part of collective goods that must be shared by both research and all subsequent consequences and innovations with the entire European population.

This data openness policy is obviously inconceivable without a European framework protecting personal data and allowing for the full valorization of this data at the European level and for the European space. The recent troubles of several social networks have shown how sensitive public opinion and users are to the use made of their personal data. This is absolutely legitimate. Because we are at a moment of change—this is not about stigmatizing or judging anyone—but we can see that, even in societies that are primarily attached to freedom and that can accept things that collective European preferences would have long refused, even in those societies, the relationship to the protection of individual data, to secrecy, is becoming a pressing issue.

Therefore, at the same time as we open data within the framework I mentioned, we need to think about the framework. One of the challenges of artificial intelligence is precisely, first and foremost, not to confuse the use of aggregated data with intrusion into personal data. This requires very precise regulation that, without hindering scientific developments—we saw this morning in the medical field, we can have a lot of useful data from a medical standpoint—preserves anonymity and what pertains to individual secrecy. What can yield the best in terms of medical innovation, as we saw this morning, can yield the worst if anonymity is not preserved, and it becomes an instrument of selection, of insurance contracts today, of access to education or employment tomorrow! And so we must think about this framework at the European level.

The General Data Protection Regulation, which will come into effect next May, which you mentioned, Mr. Commissioner, corresponds to this concern. And I want to acknowledge here the immense work that has been done. But we must go further. Because we are playing a large part of the very future of artificial intelligence. Ultimately, we are talking about control over our own lives, our implicit and explicit choices. And I say here, the European voice that posits that everyone should be able to remain in control of their data is the right one; it is the only one that can make a real development of artificial intelligence acceptable to our citizens.

It is within this framework that we must build the next steps of regulation that will allow for opening while ensuring the protection of each individual, the control of data. And then, it will be a matter of detail, innovation by innovation, practice by practice, so that this use is sincere and allows all industrialists and innovators to fully develop in a framework that is transparent and fair for every citizen.

But we must also build, at the same time as we open up this data, the framework of European sovereignty. Where will this data be processed? Who are the actors that will use it? Will the regulations they refer to be regulations that we have taken the lead on? These are real issues that will arise, that are already arising today. And at the heart of these issues lies the construction of European digital sovereignty. This is the key. In terms of taxation, in terms of regulation, in terms of law, and therefore in terms of actors, even as research is always global and perfectly immediate, its economic and financial translations always imply returning at some point to a form of territorialization. And this is what we are currently, without even saying it and without having sufficiently drawn a holistic thought on the matter, we must re-aggregate today.

We approach it specifically each time, whether we are talking about the taxation of GAFA, as they say, or the regulation of individual data or any other. But the true framework is broader! It is that global actors have emerged because their business was global. And it is global. But the consequences remain national or European, as far as we are concerned. And therefore the framework must also be able to be national, European! And the regulation we put in place, particularly on data, must be able to be national and European. And so, this is what we need to invent in the coming months and years, and you see in this regard that the strategy I am presenting today has the humility to highlight that it has an unfinished aspect, because France cannot pursue this path alone; but I want to express what spirit I will pursue in European debates.

If I had to illustrate just one sectoral asset of France in terms of data, and therefore artificial intelligence, it would precisely be that of health. In the field of health, we know that medical diagnosis, care protocols, predictive and preventive medicine are undergoing absolutely radical changes. We are moving, thanks to artificial intelligence, towards innovations that will indeed allow us to prevent pathologies, and thus drastically reduce the overall cost, which will lead to individualized medicine that will also reduce costs, improve our quality of life, our longevity, and the quality of life!

So, this transformation is absolutely radical and it will affect all sectors of medicine. And the examples are too numerous to be cited. We saw some this morning at the Curie Institute, from early tumor detection to the prevention of recurrences, gene therapy or immunotherapy, from the identification of pathologies previously considered undetectable, to the discovery of unknown molecular functions, artificial intelligence does not come to replace all the innovations made in each of these areas, but it allows for radical leaps and it clearly enables us to move towards individualization and a more predictive medicine, as I mentioned. It is therefore capable of providing therapeutic uses and fundamental well-being to the patient.

As I mentioned, in this area, we have a real advantage, which is that we have a very centralized health system, Madam Minister, with databases of exceptional richness, notably that of Health Insurance and hospitals. They are among the largest in the world and provide us with access to major scientific discoveries.

We saw this morning at the Curie Institute. Already, artificial intelligence allows for better detection of pathologies, better treatment, as we saw, of this or that tumor, better detection of cardiac arrhythmias, deeply assisting radiologists in their work and significantly reducing the deeply human medical errors that, while very minor, still exist.

Artificial intelligence will not replace either doctors or researchers, but it allows for scaling up. That is why we must fully leverage our lead in this area. That is why I want us to go even further, in order to make the most of the data available for research. We created, in 2017, the National Institute of Health Data, which manages access to the Health Insurance database under very protective conditions for the anonymity and security of patient data. And which is fully in line with the spirit I just mentioned.

I wish for us to create, on this basis, a true health data hub, a partnership structure between data producers and users, which will oversee the continuous enrichment but also the valorization of the national health data system, to eventually include all data reimbursed by Health Insurance, adding clinical data from hospitals, data from city medicine, as well as high-quality scientific and medical data created within the framework of national cohorts.

This health data hub will allow us, in a perfectly secure framework, guaranteeing anonymity and respect for each individual, to have a workspace for artificial intelligence that will accelerate the innovations I just mentioned.

All French research organizations will have their share. I mentioned earlier INRIA, CNRS, it is clear that INSERM will have an essential role in this work, and the National Institute of Health Data will continue to be the guarantor of compliance with the rules for accessing health data, because these advances will not happen if we are not able to guarantee the French complete confidentiality of their individual data.

By linking and enriching information, by working on very large volumes of data, by accelerating their opening for research purposes, I am convinced that we will invent the medicine of tomorrow, here in France. And I wish, in this regard, that the infrastructure we put in place is able to guarantee our country a fair economic and industrial return, as I mentioned, and this relates to the sovereignty I referred to earlier, and it is also important that in the criteria for opening, we have the possibility of having innovations that are deployed in France and Europe on an industrial level, to also have the creation of value that takes place there; otherwise, it would be absurd to tell our fellow citizens, "we have a wonderful resource, it will help research," very well, "but it will allow the development of extraordinary actors who will create jobs, but essentially on the other side of the world." We need to have a principle of reality here as well, and it is the sovereignty approach I mentioned, on an economic and social level that is essential.

This challenge is therefore within our reach and I deeply wish that we can lead it in the coming months. We are indeed on the brink of profound changes, and we see that the technological dimension must be supported by an adapted regulatory framework. This gives the government, obviously, an increased role in these technological revolutions, and I am, in this regard, determined not to hinder these scientific advances through insufficient anticipation and adaptation of regulation and regulation. That is the worst risk. And so, on the contrary, I wish for the State to be able to intervene, in resolute support of the actors of this transformation, but by thinking through the framework, by designing and organizing – I will return to this in a moment – the conditions for a permanently transparent and democratic debate on all the consequences that arise from it.

So, we will both open our data and the different sectors to artificial intelligence, and we will open our society – which loves this – to all the controversies related to artificial intelligence.

The third pillar of our action is obviously the regulatory and financial framework, national and European. The regulatory framework first, is that, beyond these data openings, we absolutely need to create the conditions for the emergence of these cross-cutting platforms; I just mentioned health, it is also the case for energy, defense, mobility, finance, and aeronautics. Here too, we will need to adapt the regulatory framework in France, as well as at the European level to allow the emergence of these platforms and accelerate their development, while thinking through the rules and their regulation.

In this regard, I want to take an example, to explain how the State can fully support innovation in this regulatory framework. It is the example of the autonomous car. We must recognize that, until now, France has collectively struggled to position itself among the leaders in the experimentation and development of the autonomous car, and I consider that it is an essential battle to close this gap. France is a great automotive nation, we have our historical role there.

We have largely lost out, to the benefit of our German neighbor who, at the turn of the 1980s, made more relevant technological choices and positioning of ranges than we did, this leadership. We can regain it through the autonomous vehicle and the ability to think through its framework; because the value will redeploy differently! It will shift from the car itself and the mechanical part, to the software part and the intelligence of mobility. That is why we will publish, in April, our French strategy for the autonomous vehicle, led by Anne-Marie IDRAC, whom I thank, with the ambition of positioning France at the forefront of experimentation and industrialization of the autonomous vehicle.

By the beginning of next year, we will have the legislative framework authorizing level 4 experiments which will be included in the Pacte law.

By 2022, a regulatory framework allowing the circulation of autonomous vehicles will be put in place. Ultimately, I want to establish as quickly as possible the framework for the approval of autonomous vehicles, mobilizing both public power and the private sector. Because precisely, this movement of autonomous vehicles requires a learning curve, we will finance a national experimentation program involving territories, manufacturers, and equipment suppliers. And so we will very quickly have experiments in territories that will be voluntary, defining the exceptional legal framework, and I want us to be able to bring all these national actors to the table to work on subjects such as, precisely, the test databases, the mapping.

We have, in this area, a French strategy that will therefore be deployed, but which must very quickly become a Franco-German strategy and a European strategy. And on this, the choice is in the hands of our countries. If we do as we did twenty years ago, France and Germany will divide and fight in Europe to push divergent standards. The winners will be Chinese and American, and in artificial intelligence, one does not need to be a great scholar to know that. Either we decide together – which will be my wish – to push common standards and to define together standards, common actors to define a real mapping, to define the real rules and the real creation of value. In any case, that is what I will resolutely propose.

At the European level, it is these new rules that we need to create and this deployment framework, it is at the heart of this digital strategy that we must have, which should not simply settle for an opening of our borders and the creation of a domestic market of 27, but rather the definition of common standards for 27, and the ability to create the European normalization that suits us! Because it is the only way to truly define our collective choices.

We have done it at other times in our history, we must do it today! Otherwise, we will either suffer the collective choices made by dominant private actors – and that is all their merit – across the Atlantic; or by dominant public actors in Asia, who are organizing to do so. And so, if we want to maintain the European balance and the ability to regulate that is ours, we must absolutely build this regulatory framework.

In terms of financing, we must also have a proactive approach. Artificial intelligence will be the first area of application of the 10 billion euro Innovation and Industry Fund, established at the beginning of the year. From the revenues of this Fund, co-led by the Minister of Economy and Finance and the Minister of Higher Education, Research and Innovation, we will isolate 100 million euros in the coming months for the seed funding and growth of our artificial intelligence start-ups. Beyond that, it will allocate 70 million euros per year, through the Public Investment Bank, to the emergence of so-called "deep technology" start-ups in our country.

Funding from the Future Investment Program and the Innovation and Industry Fund, amounting to 400 million euros, will be dedicated to financing innovation challenges, both upstream and downstream, or to industrial projects dedicated to artificial intelligence. In total, this will be a dedicated effort of 1.5 billion euros of public money, directly generating more than 500 million euros of additional private investments, which will be implemented to support the emergence of this major global hub for artificial intelligence.

The key, beyond these deliberate choices of public policies, is obviously that these choices and the visibility be given very early to public and collaborative research, so that it can trigger collaborative research choices and private investments, to have the multiplying effect behind these investments and public choices.

I add to this effort nearly 800 million euros that the State will invest by 2024 in nanoelectronics, an essential ingredient for the development of artificial intelligence; I know how many in this room are attached to this ecosystem.

Beyond these French choices, it is the next European financial framework that must carry this ambition for artificial intelligence, and Commissioner, you know that France will support your proposals. I believe that we need a European budget that preserves all the ambition of these historic policies, whether it is the CAP or structural funds, but that carries a new ambition particularly in terms of innovation and breakthrough innovation.

That is why I proposed, right from the speech delivered at the Sorbonne, that we establish a DARPA or the equivalent of a European DARPA and that I supported the report you presented a few months ago, at the beginning of the year, which precisely proposes funding for breakthrough innovation at the European level, the simplification of European research programs in administrative terms. All these measures in terms of ambition and organization will be fully supported by France because they are the condition for the emergence of this ecosystem and the success of the transformations I mentioned.

Beyond that, of course, the State defines the regulatory framework, finances, and Europe with it. The State must adapt itself in its organization, in its public policies to take advantage of what artificial intelligence presents and proposes in terms of education, training, and employment policy. The revolution we are facing is one that will allow, through artificial intelligence, through more predictive models, to improve the matching between supply and demand, to improve the training system and its adaptation, to individualize the capacity to train and thus allow, through our public policies, to address part of the negative externalities of technological disruptions, and that is what we need.

If we want to convince our fellow citizens that these technological disruptions are not just a vector of negative change and should not only feed fear, we must succeed in showing through these same innovations that they generate well-being for them regarding the problems they have caused. As long as we only see technological innovations that destroy jobs in traditional sectors located on the ground, it will be a source of dread.

If we can, through artificial intelligence and with these same actors and the start-ups that are emerging, show that we can very quickly predict the job losses to come in such or such region, but also those that will emerge given the qualities and qualifications present in that region, and redirect our training and continuing education system, individualize training to allow a return to employment – mechanisms that already exist technologically, we need to deploy them and translate them into our public policies – we will completely change the perception of these innovations, we will completely change the relationship, the experience that our fellow citizens have with artificial intelligence.

So, this is also the challenge posed to the State and is absolutely essential in the areas I just mentioned, and it is this dynamic that we will continue to develop and that the Secretary of State, Mounir MAHJOUBI, will precisely clarify with a national mobilization and a strategy, public policy by public policy, to draw all the consequences.

Finally, the fourth pillar of this strategy is obviously to define the ethical and political stakes of artificial intelligence. I have spoken extensively about the tremendous innovations that artificial intelligence promises, the massive support that France and, I hope, Europe, will be determined to provide, and how essential it is as an economic, scientific, and social issue. But artificial intelligence is nothing other than the project it serves, under the conditions defined by those who programmed it, based on the data provided to it, and these three conditions design, by design, if I may say so in front of you, the model of artificial intelligence that we want to see exist or rather, much more simply, the model of society that we want. And in this matter, nothing would be worse than leaving these choices to private companies or to autocratic systems. Therefore, when we talk about ethics and political choices regarding artificial intelligence, it is precisely to consider that there is an implicit behind this adventure which is the necessity to make profound choices.

As I mentioned to some of you yesterday, I see three extremely classic challenges when innovations of this type appear or great revolutions occur: we have a choice in terms of conflict between values and technique, we have an ethical tension on a geographical level, and we have a temporal tension. We find ourselves in a system and a grammar that are familiar to us in this regard, and that is what we need to think about.

The choice between value and technique is that as long as the technique serves the common good, there are few conflicts. The health innovations I mentioned, a start-up that will enable this individual predictive medicine, there will be no issue. The ethical tension and the relationship to our values arise the day when the actors who use it do not respect the collective values or the collective preferences that are ours. Therefore, allowing a start-up to use French data to enable innovations, an improvement for French patients, everyone will agree. But the day this start-up is based in a country that does not respect or does not consider itself obliged by either French law or European law and has collective preferences that are profoundly different, we will have a problem.

So this issue is that prior to these technical choices, there are value choices that we will need to make and constantly monitor. These values are those we have already begun to explore in our contemporary debates: respect for individual freedom, respect for privacy and what pertains to our own private life. But there will be other value issues that will be raised where preferences will need to be expressed, and this will need to be discussed to define the framework.

We must maintain the democratic trust in this innovation; that is the key. And so we will have value debates on this subject to determine where our preferences are and what our capacity is to articulate technique with these preferences, but we must also ensure that these technical evolutions never corrupt democratic trust because the day a President of the Republic can no longer explain to his fellow citizens the consequences of an innovation he has allowed, we will have a problem of democratic trust because citizens will say: "But then, who decides? It is no longer you, it is no longer the people we elect democratically."

To do this, we will need to ensure both transparency and loyalty in the system. Transparency means that the key is to ensure public transparency on algorithms everywhere, to make algorithms public, to ensure they are used transparently, to track their biases, not to entrust them with the monopoly of decision-making, to commit to enriching or complementing them with human decision-making. This is all that we must do in the coming months and years. This transparency requires a permanent interaction between artificial intelligence and humans and the choices that will be made; it requires ongoing debates, corrections, and it requires that from the moment we put university admissions, entry into a profession, or training behind an algorithm, the necessity to make this algorithm more democratic and thus ensure its loyalty, ensure its complete transparency, and that there can be a debate on these rules; otherwise, we delegate to the algorithm the choice between democratic priorities.

That is why I want this subject to be at the heart of the research conducted within the framework of the program coordinated by INRIA. I am fully in favor of the idea of having an open laboratory in this area and that there be, moreover, in the intersection and interdisciplinarity of the human sciences and artificial intelligence and all the hard sciences involved, the possibility to confront a reflection on the ethics of use, to gradually build the ethical codes that will accompany these essential uses. And at the heart of this reflection, we must inscribe commitments to non-discrimination based on social, ethnic, and sexual grounds, and the State, as far as it is concerned, will therefore make public by default the code of all the algorithms it is led to use, foremost among them, Madam Minister, that of Parcoursup, because I believe it is a democratic practice. I see that there are people who submitted their applications in early March, who are awaiting responses and who already want to try to see the inner workings of the algorithm to potentially thwart all strategies. More seriously, I think this is essential and it also requires a permanent dialogue, but which is at the heart of these innovations.

More broadly, I wish for a European and international reflection to be initiated on the control and certification of algorithms so that Europe becomes this space of freedom and progress resistant to both the opaque privatization of artificial intelligence and its potentially despotic use if we allow it to concentrate in public hands that are based on collective choices that are not necessarily ours, to speak in circumlocutions.

This is also a subject I want to discuss, beyond the European framework, at the international level and particularly with our Canadian friends, whom I know are very engaged on the subject. We will have the opportunity at the beginning of April to have this exchange, and I hope that we can go as far as to create a GIEC for artificial intelligence, that is to say, to truly create an independent global expertise that can measure, organize the collective and democratic debate on scientific developments in a completely autonomous manner, in a completely independent manner, and to nourish the democratic debate that we will need because one of the key issues that will arise and already exists will be the independence of research and, precisely, the discussion that will take place around not only algorithms but all reflections surrounding artificial intelligence.

I therefore believe that it is up to States, to public powers to invest in this area and to guarantee the global framework of this independence through this GIEC for artificial intelligence. It is a moral responsibility; it is also the guarantee that our democracies will not succumb, in a way, to an Orwellian syndrome where technology is no longer an instrument of freedom but a form of control, and this also requires all the interdisciplinarity I mentioned.

Articulating these values and these technical choices also means acknowledging that we need, from education and throughout education, to implement these value choices and to make them transparent, and to ensure that in our training, in our education, we are also concerned that there is no capture of these radical innovations and this artificial intelligence by a few, because today, the world we live in is a world of superstars – this is true for digital technology but even more so for artificial intelligence – where a few individuals, and I have great respect for them because it is the result of intelligence, hard work, sometimes luck, but it is there, but there is an hyperconcentration not only of the value created, but also of deep choices that are not just entrepreneurial choices, but societal and perhaps civilizational choices.

That is why I am calling for this regulation, that is why we need to define these ethical rules. But this also implies that we should not have too many similarities among the actors who are able to make these choices. And so, the actors of artificial intelligence should not be too much like what I am before you, that is to say, middle-aged white males - there need to be some, and I am not going to cut the branch on which I am sitting having been unable to change my condition - but I note that in this sector, there are many middle-aged white males trained mainly in the great American or European universities. We therefore need to mobilize more scientific training to seek out more women, to also seek more equality, more social diversity. I believe that in the challenge we face, there is a necessity for inclusivity in this training because we can see that these are critical professions where deep choices are also made by everyday actors. And so, in the strategies we must develop, we must ensure that there is not a concentration of talent, a concentration in the hands of a few.

Finally, we have to manage, as I said, the tension of geography and time. The tension of geography, as I have illustrated several times in lectures, this phenomenon is global. Research is by definition global and it must absolutely remain so. Economic actors have become global with a real concentration and sometimes situations of de facto monopoly. However, collective preferences and values that are sometimes shaken by these innovations remain national or at least regional, and that is a reality. This is the first time that to such an extent, with such speed, we have the emergence of economic and industrial innovations and such profound changes occurring without states, sometimes even circumventing states, and who have not thought through their own regulatory framework.

Therefore, this geographical tension will at some point have to be established, and I think we must establish it through the debate on artificial intelligence because it is about the capacity for sovereignty. What is sovereignty that I have mentioned several times here? It is autonomy in the strictest sense of the term, it is the ability above all to choose for yourselves the norms to which you will be subjected. And if you renounce your sovereignty in fact because you accept that a phenomenon is global and that in some way, reterritorialization comes to shake values, but it is a phenomenon that you accept not to touch or to endure, well, you accept that the rules of the game of your daily life are decided by others and thus you lose your autonomy on major issues.

I believe in autonomy and sovereignty because I believe in democratic systems - we have not invented anything better. These are systems, to borrow phrases that concern the sector, it is the most bottom-up system on Earth, democracy, because it is the people who decide. And so I believe that we must allow the democratic anchoring of this evolution by ensuring the rules of transparency and fairness, but also by ensuring that this tension between a global phenomenon and preferences that are sometimes national or regional is preserved. And so we are also entering a phase where we will need to have these democratic debates.

And finally, there is a temporal tension. I cannot finish this speech without mentioning it, it is our responsibility to all. It is moving very fast. The most talented people are already thinking about the world of tomorrow or the day after tomorrow. I have given a very small part of what I have been able to see and understand. Others are already much further ahead, but there are 20% of our population who do not have access to these technologies or who are afraid of them, and we cannot build informed democratic choices if we remain in this situation because there is a phase of democratic digestion that goes with the uses of the entire population or at least a large majority and the acceleration of our technologies.

Therefore, we will need to articulate collectively – and this is between industrial and economic actors, researchers, and state or European, international actors – we will need to find the balance in this permanent tension that will be ours on this subject where, every day, we will feel like Prometheus but where we will need to see, as RIMBAUD said, “the harsh reality to embrace, peasant.” And Prometheus ends badly when he forgets the peasant, especially in democracies. And so we will also need to live this tension, and we are not necessarily the most modern when we forget our neighbor in the hallway because we ultimately live with our neighbor in the hallway. And that is something we also need to think about in the innovations that are ours.

So, it is this framework that obviously requires massive investment in technological diffusion and in education, and it is the choice we made since last summer, from kindergarten and throughout training up to higher education, to have this choice to train, to also train in digital practices. We will also introduce training in ethics related to digital technology because we also need our future citizens to be trained in these transformations.

There you go, Ladies and Gentlemen, I have already been very long but it is a continent that you have proposed to me, with you, to embrace. Ultimately, we return to a new very Cartesian stage of this ability to be master and possessor of nature, and it is in this balance that we must always play and build our action. The strategy I have just presented, which stems from the work that many of you here in this amphitheater have built, will constantly need to be reviewed. It is not a vertical and fixed object because in six months, a year, it will need to be adapted, transformed, shaken up. But I believe that to date, these four pillars I have presented are the four deep convictions that will guide our choices, our investments, the spirit of our laws, and the commitment of France and, I hope very deeply, the commitment of Europe to successfully undertake this adventure of artificial intelligence.

So, you have understood, I want France to be one of the leaders in this artificial intelligence, I want Europe to be one of the leaders in this artificial intelligence. We have the means and we will create the conditions, first, because I think it is good for our country, then because I think it creates opportunities, finally because I think there is no chance of pretending to control the effects or have a say on such or such perverse effect if we have missed the battle at the beginning, that is to say, to be an actor in this adventure.

So, I will do everything to ensure that we succeed, that we are in excellence in training, research, and economic and industrial excellence in this area, and I will do everything to ensure that we are also the place where we think cooperatively, openly, among all sciences, between our citizens and the actors of this transformation so that we think about artificial intelligence, its consequences, its place in society because that will be the best way to reflect together on the society we want to build. Thank you for your commitment and, as you have understood, you can count on me – I say this here without any innocence – to build the true renaissance that Europe needs.

Thank you.