Making Europe a leader in AI

Show video

Good afternoon. Um, i'm vanki ramakrishnan. The president of the royal society. Which is the independent. Science academy, of the uk. Thank you for joining me in this international, event. Where i will be speaking, to antoine, patek. The chairman, and the chief executive. Officer. Of the santro, nationale. De, la research, scientifique. Or cnrs. As we, call it. And. Professor, martin, stratman. The president of the max planck society. And the theme of this event, is on making, europe. A leader, in artificial, intelligence. During this discussion. We will have contributions. From some ai, specialists. And end with some questions, from the. Audience. As a self-governing. International, fellowship, the royal society, is committed, to fostering. International, collaboration. And convening. Colleagues, from around, europe, to discuss a global, issue like ai. Really underscores. The importance. And necessity. Of cross-border. And international, collaboration. To exploring, new ideas. Testing, theories. And comparing, and contrasting. Different, scientific. Perspectives. Currently. The dominant, players, in the development, of ai, and digital, technologies. Are the us, and china. And it is often said that europe, plays, a minor, role. However. There are several, examples. Of, really. Groundbreaking. And revolutionary. Digital, innovations. That came out of europe. For example. Europe, invented, the world wide web. Which actually. Gave the internet, its power, and reach. And really made. A. Drove an explosive. Growth. Of the internet. It was initially, developed, for the exchange of information, between, physicists. At cern. But then, it was developed. And scaled. In the u.s. And it also shows you how. Discoveries. Made in one context, can have really, broad, applications. That, are not entirely, anticipated. Europe also, pioneered. Multimedia. Communication. Platforms. We all know about skype. Skype was actually created. A technology. Underpinning, skype was created in a collaboration. Between. Computer, scientists, in sweden. Denmark. And estonia. The world wide web, and indeed, much of the internet. Is actually driven, by, servers. That. Run under the linux, operating, system, and linux. Itself. Was. Created, by. Finland's. Linus, torvalds. So, another. Example. Of. European, contributions. That have revolutionized. The growth of the internet.

And Computer. Technology. And europe is also at the leading edge. Of developing, the hardware. Supporting, advances. Including, advances, in ai. And it's for that reason that nvidia. Which produces. The graphics, chips that. Run. A lot of the. Machines. For ai. Is actually acquiring. The uk, based, chip designer, arm. For 40 billion dollars. Um i should point out the uk is home to, many, cutting-edge, startups. That are much sought after, for example. Deep mind, founded, by, demis hasabis. Was, a fellow of the royal society. Is a good example, of a highly, innovative. European, ai, startup. That grew and attracted, the interest, of tech giants. And. It is now of course. A. Part of google, although a fairly. Autonomous. Part i should add and still based in. London. Another, area, where europe has shown leadership. Is the governance, of the use of data, and ai. The use of. Data, and ai. Has a number of problems. Both ethical. And. Questions, of privacy. And security. And europe has pioneered, bold, regulatory. Frameworks. That can influence, the use of data, and ai. While data is global. Gdpr. General data protection, regulation. Has shown it is possible, to legislate. Technologies. At the territorial. Level. Cdpr. Has inspired. Changes, around the world. California. Followed suit, and large, technology, companies. Have increased, their focus on protecting. Privacy. Tech giants are looking to europe, to clarify, the rules of the game. And companies, responsibilities. Regulation. Can often be seen as a break, on, innovation. But done well it can actually, boost the development, of ethical. And trustworthy. Technologies. And approaches. And in fact, enhance, innovation. For example, technologies. Known as privacy, enhancing, technologies. Or pets. Could be further developed, and help use data and ai, safely. So europe has an opportunity. To spearhead, the development, and use of such technologies. And indeed, to set standards. Which could be important, for the future. Governance, is not just about regulation. A number of european, countries have set up commissions, to scrutinize. The ethics, of data, use, and ai. And for example, recommendations. From the royal society, and british academy. Review. On data management. And, use. Governance, in the 21st, century. Led to the creation, of the uk, government, center for data ethics. And innovation. One of the roles of such stewardship, bodies is to ensure. That data-driven. Technologies. Are developed. And, deployed. In a way that works for everyone. Across, europe there are examples, of how public debates. About the place of science, and society. Can shape the development, of research. A well-founded. Public debate. Based on continued. Engagement. Between, ai research, and the public. Will be important, in shaping the development. Of ai, technologies. To sum up. Clearly, europe is capable, of producing. Cutting-edge, innovation, it has done that in the past. Several decades. And it will continue, to do so in ai. And, as well as playing a leading role in shaping, public debates. And policy. A major, challenge. In europe now, is to ensure. That it not only develops, technology. But allows, the growth. And the scale. Of technology.

In Other words escape. Scaling, it up, uh to. Reach sort of global, levels. Of impact. In the first part of this event. I'm going to invite, antoine, petit, and martin, stratman. To. Provide some opening remarks. And i'd be especially, interested, in hearing their views on the following, questions. How competitive. Is europe, in ai, research, and innovation. And what are europe's, strengths. The royal society, conducted, a number of studies, on the benefits, and levels of uptake, of ai and data science. And found that data, is a key, challenge, for the uk, to realize. The potential. Of ai. We have also found significant. Growth, and demand, for skilled, data, scientists. Have cnrs. And max planck, found the same, in france, and germany. Or what are europe's, weaknesses. In ai and how can we address this. Work by the royal society. The leopoldina. And aleya, the network. Of all european, academies. And by others has, identified. That europe is in a unique position. To lead on the development, of trusted. And ethical, ai. How has europe demonstrated. Leadership, in this area so far. And what could it do further. Ai, is said to have the potential. To tackle. Major, societal. Changes. Challenges. Such as climate change. The response, to the covet, pandemic. Or broader implications. For health care. How are our political. Leaders, defining. Their ambitions, for ai. Across, europe. And how can we as the science community. Inform, these ambitions. We face a number of challenges. At the moment, the pandemic. Broader, political, issues such as the. Future, relationship. Between the us. Uh between the uk, and the eu. As well as between, us. And the u.s, and the. Asia and the rest of the world. At the same time. The eu is concluding. Discussions. On the next research, program, horizon, europe. Given these challenges. How can we work together. To make europe a leading destination. For ai. What we need to see in terms of research, and mobility. Regulation. And funding. To facilitate. This. So i'd now like to, hand over. To antoine, petit, the chairman, and. Chief executive, officer, of. Cnrs. And, a computer, scientist, himself. So, over to you antoine. Thank you very much uh vengeance. Really a pleasure to share. This event with you. So. I am convinced. That. We are only, in the, infancy. Of ai, i can say. And so that almost, everything. Remains. Remains, to be invented. But of course, the race, has started. And the question is whether, the, european, countries, as you mentioned, or even better, whether europe. Has a chance to win it. And the race is not lost. Clearly, but we have to make an effort. Right now. And to be a leader. Or one. Of the leader. I think that several. Ingredients.

In Some sense, are needed. I would like to mention, five of them which appear to me as, the most, important. First of one and i will comment but briefly the list first of one, is. The question of data that you mentioned, already. The second one is the equation of the computing, power. The third one is a question of skills, and talents, and in fact human, intelligence. But the fourth one is also the problem, of the relations. With private sectors. And last but not least. The, fifth one is the equation of the accessibility. And it goes back to trust. And fairness, as you mentioned. Concerning, data. You say that you said that data is a big challenge for uk. From my point of view data is a big challenge for europe. And, and i want to include uk and europe. Of course. Uh. I mean by that that i think that each of our country, in fact is too small. Uh to to to deal with. Only its own data. And that we have to share, perhaps we have even some times to share at the international. Level. And, the recent, crisis. Shows that probably, we should have, share more, even with, our china, china, partners, and. And american, partners, but at least in europe, we have to make an effort. And i know that it's not easy. If i, look at the example, of france, even, inside, france we have sometimes, to share data. And so, it's clearly. An issue. Uh. And it's, of course very important. That, this data, or. Cross-referencing. As much as. Possible. And we all know that gaffa. And, their urgent. Counterparts. Have already, a lot of data, but nevertheless. I think that we still have to work on that subject. Also on data. Between, business, business to business, data. And. And the. 5g, revolution, will produce. Again more data. But clearly for me something that we should, share at the european, level. The second point concerning. Computing, power i will be very fast because i have the impression. Perhaps wrong but i don't, think that, it's not the main problem, right now in europe. So third one is very important, it concerns, human intelligence. And, and, you mentioned, already this question of talents. I think we me. If we want to have a, strong europe, we need to have, people, educated. In ai, at different, levels, we need we need of course a phd.

Laureate, And phd. As if so we need to have, education, at the phd level, but we also clearly need graduate, and undergraduate. Specialists. Of data and ai. To work in industry. So that's clearly a question of education, so what we could do as a european, level. I think we have to make an effort. To uh, to. Attract, and to retain, talents. We all know that the competition. Is, worldwide. And for instance, i am in the board with in charge to to look at the chairs, of ai, in canada. And, they. Have decided, to have 60, such shares, until now and, half of them. Come from outside, canada. With really outstanding. Uh. Profile. And that's something we should push. At the european level. Perhaps, using, tools like the erc. Which is so important, for for europe. But clearly, it's a it's a. An issue. And we have to be optimistic. In the sense that a, lot of, ai, is better. Based on mathematics. And computer, science, as we all know. And these two domains, are clearly domains, in which europe, has. Already. As a, for always, very good, performance, at the international, level. But it's also, important, i think that we we, we push, to have. Ai, and other sciences. As you said i'm ceo of center life seniors, covers, all fields of science, and i am very, impressed. By the first results. Of ai. Applied, to other, subjects, for instance chemistry. And ai. Or, the questions, of data, in astrophysics. Or particle. Or particle, physics, it's a they really. It's really rich. Amazing, new techniques. And so we need. Talents, and human resources. Of course in ai, in pure ais, i can say, but we need also people with some kind of of double. Competencies. In ai, and you know in another field. So, the fourth condition, from my point of view, is the what i will call the. Private, ecosystem. And its relations. With, uh, with uh. With academic, world. So, it's it's a program in europe we do not have very, big players. As it has already been mentioned. So we have, quite promising, startups. But one of the problem as you you said bankies, very often. These startups. Are, after, some, some years, or some months even they are bought by american, or asian capital. And uh and, which is not so bad but unfortunately. The converse, is not true, uh, and so that. Creates a. An unbalanced. Situation. Which is not. Not from, my point of view very, very good. But we have big companies. In europe. And. I think that one issue is that these big companies. Must, make. The ai, gamble. And i'm not sure that all of them. Did it, already. And uh. Because clearly, this ai, gumball, is not, an evolution. But clearly a revolution. And this revolution. Requires. Agility. And questioning. And that's not so clear. That, a lot of big, existing, companies, are able to do that. Now. And we have also to develop, interactions. Between, private sector, and academic, world. I was mentioning, the canadian, shares, and i am very impressed by the fact that. Almost, almost, all our colleagues. Most distinguished, colleagues. Have some kind of double, appointment. Between a canadian, university. And a big american, company. And this is a system, which, does not really exist. At least in france. And uh. With also it's a system with some drawbacks, clearly. But i think that we have to to think to that it's it's clearly, facilitates. The interactions. Between private sector, and academic. Uh academic, world. And so on last but not least i would like to. To as you said thank you to to insist, on the fact that.

The Question of acceptability. For society. Is a key issue. And from my point of view this includes a question of fairness. Of explainability. Of ethics. And. From a scientific, point of view. It means that we have to, to, also, to deal, with humanities. And social sciences. So of course ai. Is clearly, a lot of mathematics. And a lot of computer, science. But, the question of acceptability. Of explainability. We require. To have also, specialists. Of humanities. And social science. Which will help us also to understand. In some sense the behavior, of people. I mean by that that, when you, when you see a doctor, and if he gives you a diagnosis. Few people, in fact will ask questions about why this diagnosis. But when an, iai-based. Algorithms, propose a diagnosis, it's very different, people, require, explanation. And this, and there is, it's normal, but it's, interesting, to see why in some situations. You you you need. Explanations. And while, in some other situations, you don't, need this, situation, this, explanation. If you if you take a plane for instance we know that a plane is derived now by but not by the pilot, by a computer. And nobody, cares, and and. So you don't ask any, any explanation. When you when you go to. To a plane and. If with ai's. The situation, is very different, and we saw it in particular, in france with the, application. Called stop covida. Which has not been accepted, by by. By the population. And so this, equation of ethics, and. Seems, from my point of view absolutely, crucial. It's important, to to mention, that we do not have the same ethics. Than the american, people, or china people. And their ethics, is not better or worse than ours, but it's different. But i think that in europe, more or less we share, some kind of same ethics. And so probably, we have to to, pay attention, to this and, as you you said ranking. Perhaps, it's this, this. Ethics, ai. Which is uh. The fill of the domain. In which, europe. Could take. Some kind, of leadership. Or, at least to not let the leaderships. To, china, or asian countries. And, america. Thank you. Thank you very much antoine. Those were, very insightful. Remarks. And. I'd now like to welcome. Martin stratman. And, if you could. Uh give us your perspective. On ai in europe. Yeah first of all thank you thank you for inviting me today for this meeting. Uh i'm a chemist by education, so i'm not a specialist, of the field. But, i might try my best to give, an input, to the discussion, of today. Um, also. As part of my activities, as president of smart sponge society. I would like to spend a few minutes, in the very beginning on the science. Of. The situation, of the science in europe in general, and then i would like to focus. On today's topic. Let me start with you comments on the general situation. Of the science, science landscape, in europe i think europe these days, faces, a number of challenges. One of them some of them are specific, for europe, others, are more general nature. We all face the immediate consequence, of climate change. We see a new kind of industrial, and we even call it, post-industrial. Revolution. Which will transform. Major, parts, of the european, industry, in ai as part of this. We just experienced a pandemic. Which already today, dramatically. Accelerates. This transformation. Process. Giving a boost. To digital processes, and business. And obviously. Digital industries. Like online retail for example is on the winning, side, and last but not least as a specific topic we have the impact of brexit. Which may. Unfortunately, if it runs unfortunately. In the end. Uh come up with additional borders in europe. And in fact also the industrial, performance, of europe in general. We all know i think with the exception of brexit. That all the issues i mentioned before, can only be dealt, with by science and science, is not only a toolbox. I mean a toolbox to handle specific, immediate, problems. But science. Serves as an underlying, fundament, of knowledge. As basis for decision-making. Processes. And for the as such i think science is a fundamental. Importance. For the well-being. Of european. Countries. And its citizens. Now how about the status of science in europe if you just compare, the worldwide. Top one percent, articles, and reviews, so the top papers which are going to be published. The eu, 28, which includes of course britain. Is on par with u.s, and i think that's not bad at all, and if you just see the noble. Prizes which are given. Again. Europe is very much on par with great britain. What is behind these numbers.

I Think these numbers behind these numbers is are two aspects, first of all, there's great science, being done. In individual, institutions, in europe. But in addition. We have what we call intergovernmental. Agencies. Which stand for the highest, reputation. In science i come to this later also in, the, aspect of ai. We have cern for example particle physics we have the, azo, the. Southern, observatory. For astrophysics. The aesa for space science and the amber. For molecular, biology. We have european-based. Large, scientific, infrastructures. The s3 roadmap. We have european grants. Which support. Excellence. In science crc. We have guaranteed, mobility, in europe, i i. Would like to mention the mercury, mari creek program. Which is based on common. Common, social theory, standards. And of course we have very strong collaboration, among scientists. Which is strongly supported, by the european. Science program the, framework program. Including, also widening aspects so many, much has been achieved in the past. But there are some pressing, issues. Which you also have to solve. And among them is. That to my feelings the science and innovation, budget, within the european, framework, is too small, we will see that will, also affect, ai. The scientific, performance, gap. Between east and west and northern's house. Sometimes, hinders. Common agendas. And of course a brexit, process a threat to the, european koreans, also in science. Ai, is one example. I think why splitting, up europe, is a really. Dangerous, process. We know that international, competition, is fierce. And each country in europe. On its own. Would be much less competitive. In comparison. To cs or china. Second. Ai based business depends, to the best of my knowledge strongly, on scaling. And i think we will only be competitive, on a european, level. Not. As individual. Entities. And last but not least data, are essential, for ai. And, to me it's obvious and you have mentioned this also before that europe will have to provide. A safe data space for industry. And science. As one. Cornerstone. Of future ai. Now let me. Have some words on today's focus. On ai, on itself. To me ai is a fascinating, field of science that's first. But it has also an enormous, impact on our economy. Ai. And related, technologies. Will boost, and i feel transform. Our economy. Many jobs will be lost in people's fields already. But many new ones will be created. I just would like to mention. The human, robot collaboration. Some call it cobots. Is on the rise. And europe, should not miss out to profit. From this kind of new steam engine. That is ai. I think this process. Will be transformative. For major part, of our successful. European, industry. Now to fulfill. The promise. And as one example, also to stimulate, the strong european, robotics, sector it's only one example. And, japan, and europe are the key players in this market. It will be essential, to transfer the strengths. In industrial, robotics, into the aih. And i think it's only one example. And we all know that china. Is gaining, ground. But. Of course ai technology. Not only. Brings genuine, products. Product innovation. To the robotics, sector. And make make robotic, users that we say more efficient. It also, makes possible, the rise. Of completely. New type of companies. That provide platform, technologies. Which are applicable. Across, very many different sectors. And i feel that's one conclusion, i have that in sectors where europe is really strong. There is real potential. Just think of the energy sector. The chemical industry. The catalysis. Development, or the development, of new cuts catalysts. Health. All these big items. Are much more complex, let me say than just, search engines as we can see them, and i think they're here, the future is still open the eye will play a major role and. Europe. May have a major. Say in these fields. For all of this to me it's obvious that top research, is needed, that's i think, my core message. Platform, spanning cortec, and innovative, robots, for example. Basically, arise directly. From the best science. As it's being done in the best academic, institutions, in their spin-offs. And at the heart of all of this is people. People who advance, modern ai, for example like machine learning. I think, suddenly we all agree on that such people. As a breeding ground. For all living, ai. Ecosystems. They virtually, open up spaces, of possibilities. Which others. Can then explore more deeply. In science, and also in industry. So i it to me it's obvious that the strong and vital to european, research, community. Is absolutely. Vital. For providing. These kinds of people. And so we also may ask. What is the situation. And you ask this um. Venky what is the situation. Of an eye in europe. Now you can have different statistics, on this i recently had a look at glance, on the.

Uosad. Database. And then you get an idea. Where the top by eye signs comes from. Based for example. On an analysis, of the top one percent of, confidence contribution, in this field. And you can see at the moment that europe is actually more or less on par with china, but china's showing showing. Fast. Um. Fast speed and fast increase in their publications. The eo27. Are even better off. In terms for example of top one percent journal articles, or top one percent patents. But. At the moment yes. Still, plays in a completely, different league, than all the others and so there's lots of room for us to improve. But we also have to keep in mind. We are not lost. We do have strong research, sites. In all our european countries in france for example the uk or germany and many others. In france we have indriya. We have the crs, we have the economic. Superior. As really strong players. Paris, is certainly, the biggest hub in france. Also if you take the compasses, in southwest. Into, consideration, university, parisia, clay for example. In germany, the max blanc society, has strong activities, in ai. Particularly, in tubingen and stuttgart. Other sites are munich, the tesla university, of munich or berlin. In uk, many sites, oxford, cambridge, london, edinburgh. And overall in europe we have zurich, amsterdam. And many many, places more so in a nutshell. Europe has a lot to offer. Even in the field of ai. But competition, is fierce. And europe is lagging behind. Top sides. For example as comparison, to the s. And in particular. Europe does not have companies, like google amazon, and so on. And so my conclusion, is, we should not and we cannot just copy our, strategies. To be successful. We have to come up with our own. European. Concepts. Views. And visions. Now what to do. First of all. People. We have to support. And attract. Truly outstanding. Brilliant, individuals, i think that's, that's in, that's a call that's nucleus. To this, to me. The erc, program of the european, community, is essential. And for that reason we really, i'm worried that really this program, is, financially. Stepping, down, we should increase, this program, we should we should make it much more efficient. We have many more people who would like to join. European. Science, sites. In particular, as these days the us. Is not doing that well. We need also excellent, theoretical, training, in mars and computer, science, also that has been mentioned, and here eastern europe, has a lot to offer.

Because, Mars, for them was always, extremely, important. So people is one thing, the second element is networks. We must create, a stable, and powerful, network. With our existing. Outstanding. Institutions. The ones i mentioned before. And i think that has to be done bottom up. And probably, and that we may discuss, probably. It can only be achieved, in parallel. Not in part but in parallel. To the, eu framework, program. Since regional, funding. Must not play a role here, we only should look for excellence, here. And other. Other funding programs for example. The or the goal of other planning programs, is. Is also to support, let me say. Um. Parts of europe which are not doing that well because it's the widening programs but here i think we should really concentrate, on outstanding, institutions, which do exist, which have a track record in ai. I think if you fail. I think we will fail. If you mix up. Individual. Regional, interests. With the aim of identifying. And promoting, strong partners. Now how can we do so. Forming these kind of functioning. Networks. I think, europe. Has success, stories, in this field to offer and one is the amber. Embl. We all know, how to get it how to get this working we have seen this we have we know how the emblem works the embl. And we know that that's a highly successful. Structure. For me in the field and this is my personal view, for me in the field of ai and machine learning, the alice. Initiative. A european, network, of excellence, which is based on top-notch. Fundamental, science, is a real chance here. Only those. Who belong to the top groups in europe, can join this initiative. We should promote, alice. At a european, level. Through a treaty. Between, governments, so my conclusion, is, let's transform, the alice initiative. To some kind of modern amber, fosic, focusing, on basic science in europe in the field of ai. And i have to say i'm really impressed. By the speed. The alice initiative, has taken off. Recently. I think these days nearly 30 units. Have joined this initiative. So the first was people the second was networks. The third one i think is infrastructures. I feel that we need a dedicated. Computer, infrastructure, in europe. As support, for european. Ai. And i personally also feel that looking at the intermediate, and long-term. Time, long-term. Timelines. Quantum, computing. Is of real interest. So i would say that at the moment that ai, and quantum, computing.

Are Able to couple this in the years to come. Could be a real game changer. Europe, has a lot to offer here. And it's a real chance virtual, which we shouldn't miss, to bring these two different fields together. We also have to ensure the availability, of data we have mentioned this before. Data and europe, come from science. They come from industrial, partners. They come from medicine. For example, gaia x is one step in that direction. So, as a huge advantage, i see that we can and should combine, hardware. And software in europe that's also very specific, ai and smart robotics, for example, we have beautiful. Industries. Uh in the hardware field and we should combine this with modern software now coming to an end. I think we have no time to lose we have really to act. Um. We, otherwise, i think we will lose. Against, um, institutions, for example in the us and according to the best of my knowledge, the most. The most. Famous. Or the best performing. Computer, science, um. Place in europe is the adh, eth, these days, and it ranks only on place, 28. On the world list if i'm correct so i think there's really lots lots of things to do, my recommendation. First have a clear focus, on proven excellence. To concentrate, on this and do not delude things. Second. Concentrate, on future, aspects of the eye, machine learning, explainably. I, self supervise, learning. Etc, etc. Third. World work hard on bottlenecks. Such as hardware. And real world data. Force, expand, and strengthen, alice, as an intergovernmental. Institution. So that europe as a whole becomes competitive. And last but not least. Make sure that ai made in europe. Is. A superior, offer to people and companies. In ensuring, that it's based on clear standards. Trust. Transparency. And also as has been mentioned. Obvious. Ethical standards. Thank you. Thank you very much martin. And thank you both for your very insightful, comments. Uh before i ask uh, some of our other. Uh contributors. I just want to, say i very much agree. Uh with the importance. Of. Trying to manage, data on a europe-wide. Scale, rather than in a fragmented, way. And the importance. Of. Building, institutions. On a europe-wide. Scale and infrastructure. On a europe-wide. Scale. Because i fear that's the only, way that we'll be able to. Compete. In the global, market. And. You mentioned the us and china. I do. One thing i might say is that china, today, is not. Where china, will be. In five years in terms of its trajectory. China is actually, accelerating. Enormously. And part of the reason, is that. Data, availability. In china. Is different, from data availability. Here, etc. Now of course we, would not necessarily. Want to copy that sort of model. And i think europe, offers, a. Kind of third its own unique, way. On. Access, to data, while preserving. Appropriate, safeguards. And as you say. Producing, a superior. Quality. Of ai that can be trusted, that's transparent. Et cetera so thank you both. Uh for your comments. So i'd now like to. Invite. Marta, fiatkovska. Who's a professor, of computing, systems. At the university, of oxford. So martha we, welcome your comments, at this point. Good afternoon. And thank you. Venky, for this introduction. Into my question. Um. Algorithmic. Decision, making. Is at the heart of a.i. But we have seen a fair share of successes. But also, unfortunately. Some well publicized. Failures. We've seen, security. Risks. Privacy, leaks. Poor, implementation. Of decisions. Unfairness. Etc. Now, at the same time, algorithmic. Decision, making heavily, depends. On data, being, widely. Available. And across, borders. But also protected. Nobody, wants to see the job applications. Or, medical. Records, leaked to the internet. Well-publicized. Failures. Unfortunately. Affect, acceptance. Of the technology. By the public. And they agreement. To release, data for general, use. And this creates, a negative, feedback. Loop. So my, questions. Are. What, are the main, challenges. In ensuring. Data, sharing. Data, availability. Including, across, borders. And consequently. Acceptance. Of algorithmic. Decision. Making, by the general, public. How can they be overcome. And does, europe. Have an advantage. And if so what is it. In particular. Is gdpr. Such an advantage. Perhaps, i can quickly respond before going to the others, i think europe, does have.

An Advantage. Because, it has always. Had, a good. Framework. For regulation. And trying to balance. Regulation. With the growth of. Businesses. And while supporting, business. And the gdpr. Is just one example. Of that. But you've also seen, in many cases the eu has. Acted, decisively. To break up monopolies. We saw that with microsoft. Earlier, with google. And this is going to be an increasing. Problem. And i so i think europe, offers. What i would call a, well-regulated. Marketplace. Where. Ethical, principles. And privacy. Are considered. More important. Than, perhaps in other parts of the world. But perhaps. Some. One of the two others would like to answer, antoine, would you. Uh like to go first. Yes, pleasure. You're a bit optimistic, from my point of view venky. I think that regulation. Is good but sometimes. It can be also, uh. Something which not which is not so good. I mean by that that uh, to share data. Is as i said at the very beginning it's very important. And sometimes. I'm. Sorry to say that i think that we have too many regulations. And so i think we have to find a good balance, because if we if we had no regulations. Of course. People will have no not any trust, in ai, in in. Labs. Using, ais, and that that's not possible, of course in a, democracy. Like europe. But on the other way i think that if we led the floors, to lawyers. And then. The the race will be finished. And we will have not decided. Exactly. What, is accepted, or not. And, if i look at the example, of gdp. It's, quite a good example, because i think at the very beginning, it was also. In order, to put some limitations. On big companies. And now, it's, sometimes. More, annoying. For for even for public, labs, than for large companies. So. I think it's, it's a, very very, difficult, question martha. But of course a very good question, and i think that, the key point would be to find a good balance. And it means that at the european. European, level, we have. The decision, uh, the people who take the decision. Have to, to understand. In some sense. That the share of data. Is very important. So regulation. Yes but, but we have to go fast. Thank you antoine. Martin do you have any comments on that. Yeah, two comments mainly uh first of all this is of course uh in the core of what we discussed i think the trust of data i think at the moment we see a situation, where all these data, mainly belong to a few companies they have a kind of multiple, structure. They, um use this data, on for their own business models get extremely. Wealthy, um due to use of data. And there's of course a lack of trust of many, in the population, that we see on the data being, gathered and being used and saved by the big tech companies, we see these days. Um i give you one example for a huge data set which i found. Absolutely amazing and this is the danish, bio data center in the danish biodata, center. All the danish people from their birth on, they give a num heaven number and all the bio data, are being stored there including by the way like physical data i mean blood samples, et cetera, so all the, tests you undergo, in your life. In whole denmark, have been stored in one, place. Highly. Highly. Valuable data i have to say. Now how is this possible. It's possible there is obviously a trust, between the population. And the government that this is not misused, second there are strong relations, how to use this data space because otherwise it would be a total mess, and you could imagine that if you if you think in modern let me say, drug. Therapies. It's a huge data space you can use because the therapies. And blood samples, and all this stuff, is included, for millions and millions and millions of people. What the what i was saying for this trust is first important so we have to work very hard on trustful. Data storage, and i know, that in some states like in the northern parts of europe, denmark, sweden, this trust does exist. In some other, states, it does not exist but we also have to know in the united states it doesn't exist at all so we have an advantage, i think with respect to the united states i think china. I don't think the population, has any trust in what is going on there. The second thing is second element i want to mention in germany. We we have not developed what we call a national data, uh. Infrastructure, for science. So more and more are seeing all the scientific, data, are pulled, again it's a trustful, data space highly valuable data. Again, it's a it's a it's an it's a system which is trustful, because it's let me see enclosed. Also by rules and regulations, and some rules and regulations, you have to have, otherwise you do not put all your science data, in the common data space and i think that has to expand, in europe. And third. And that is a consequence, of the competition. The company's. Experience. There has a project between germany and france which is called gaia x which is a data storage, room, for industry. Again i think german and french and i mean if you all the others they all see that they, lose competitive, edge with respect to the us.

So They have to team up they are competing, let me say in certain fields so they have to have a trustful. Data space and gaia x is one of those attempts, to achieve this so i see this these, things to happen. And they happen because. Europe feels that they have to do something, in the startup space, and they cannot just rely on what is going in the us. So i think this combination, of regulation. Trust. Trustful, governments, which i think are essential. Can be. Can be a nucleus, let me say of of, european sovereignty. In data, the data. Are there of course in europe and we should not lose them all, to the us, or to any other state. Thank you very much, uh martin. So i'd, not like to introduce. Stefan, mullatt. An academician. And professor, at the, college, de france. And, uh. So welcome, and we look forward to your comments. Thank you very much thank you so, uh, i would like to, make. Comments. Uh, first about fundamental, research in ai. And ask, a question which is probably the, most difficult, one concerning, talents. And, european. Sovereignty. So, i think it's important, to, first realize, that what limits. Applicability. Of ai, and machine learning. In industry. Is not just technology. It has been mentioned, but i would like to emphasize. It's also about fiability. Robustness. Biases. Explicability. And the reason, why these problems, are so difficult. Is that fundamentally. We don't, understand. How these machines. How this, algorithm. Works. In the sense that we don't understand, the mathematics. And there there are, fundamental. Pluridisciplinary. Issues. Basically, the problems, are in some sense quite similar, to, physics, where you have billions, of interacting. Variables, provided, by. Measurements. And, you are extracting. Some microscopic. Measurements. Decisions. And these processes, which are similar. In material, sciences, or statistical. Physics, are not understood. And so i would like here to emphasize, what has been said before. Fundamental, research, at that stage of ai, is very important. And we can only, i think, go forward, and support, erc. Which there i think have been, absolutely, great, for. Research, in uh, in europe in that domain. Now the most, difficult, problem where i would like to ask. The questions, concerns. Talents. And.

Sovereignty. So in france. We have, great startups. Very lively, dynamics. But. As, it has also been mentioned for england they are being both, and in particular. By, a u.s, and asian company. So, what's different, from the other sectors. Like the one mentioned, by martin. Concerning. Mechanics. Chemistry, and so on. It's the situation, we have to set, of industrial, domination. There is a brain drain. But this brain drain is not towards, the u.s. The brain drain, is really, towards. The gafam. Lab for example, there are many in paris, and in france. Of course hue way is also, installing, two labs now in paris. And these are great research, labs with computational. Facilities. Data, availability. Top researchers. With excellent. Salaries. There i think, that we really, need a european, position. To understand. How to tackle these issues. Relatively, to us, and of course. China, and i i'd like to spend. One minutes, to explain, the difficulties. It's important, i think. To understand, the mechanics. Or the mechanism. Of what's happening. So. These uh gafam, companies. And some asians, attract, really the best talents, and the reason. Is simple, the salary, is essentially. A factor, five. Uh. And that creates, deficit, for public research. For teaching. And of course as. It was, mentioned by antoine, teaching, and i believe it is extremely, important, now. For educating, the next generation, to this technology. But it's also, a deficit. For the european, large companies, which have a real difficulty, to heart. The best researchers. In this, near. Monopolistic. Situation. Now. These companies. Have presence, in all european. Institutions, and universities. They all need. Money, computing, powers. And they compete between themselves, to get it so. Of course. We all need, this money for carrying, our research. And there is a consequence, of that is that of course, very good students. Prestigious. Image, they obtain, out of that. And i must say also. A. Very strong research, link it's. Your friends, my friends, my colleagues, are, in these great research, labs, how can i say no to collaborate. Even if they pay. Some of the grants, even refusing, would be, insulting. So. The situation, i think is interesting if we take for example, elise, which, i believe is an absolutely, great initiative. And i'm sure neria, will speak about it. But if you look at the first page. Of ellis. The first. Support, comes, from, google, james dean, which says it's great. And is also saying that ai, is the biggest, growth, factor financially. For google. Then comes, as a support, facebook. Then comes. Amazon. Then comes. Qualcomm. The difficulties. Are great and i, understand, perfectly, why ellis did that in the same way that all these lab. Do it. And. What i'm afraid is that in time of kovid, where. A lot of companies, have difficulty, european, companies. And where these. Companies, are on the other side growing very, strongly. The situation. Is, probably. Going to be even more difficult. So, my questions. Is for this great research, institution, that you are representing. How do you think we can deal with it and again i think we need to deal with it at a european, level. Individual. We can't resist. Local, institution, countries. What do you think can be done. Um. Thank you very much, stefan, for those, remarks, and your questions. I noticed we're quite a bit behind, time, so what i would like to do is invite. Uh nuria, oliver, to, also. Add her comments, and then we can perhaps respond, to both of you. Uh together. Thank you, thank you thank you that's great because i was i wanted to answer, uh to stephan's, last comment, so, just clarification. On the ali's website. We have a, few tens, of supporting. Uh entities, institutions. And none of them, uh. Doesn't mean that they provide, financial, support, it just means that they actually support the vision, for ellis, so, just because i thought maybe, uh there was a confusion, whether, they were sponsoring, alice in any way, we have another program which is the sponsoring, program. But anyways, uh. It's a pleasure to be here and i as he has already been mentioned, i am here, representing. Ellis. Which is the european, laboratory, for learning and intelligence, systems and i think it resonates. With a lot of the, comments, that have been made so far. So elise is a european. Ai network, of excellence. And he focuses, on fundamental. Research. On technical, innovation. And on having. Positive, societal. And economic, impact, through, modern, ai. And what we call by modern ai, is the approach, to build, artificial, intelligence. That is, data-driven. Based on bottom-up. Methods, having machine learning at its core, so we've talked a lot about data. And, you know the ai methods, that are trained with these massive sums of data, are sort of like the methods within this modern ai. Field. So the main. Characteristics. Is in a similar way to human intelligence. These learning approaches, to ai. Don't need to be, explicitly. Programmed but they are constantly, learning from experience, not required learning from data.

And As benky explained, in the introduction. A lot of the, fundamental, ideas not only in computer, science more broadly speaking, but also in in in this field, were developed, in europe and were developed by european, researchers. But as everyone, has mentioned. Europe is facing, very serious. Challenges. When it comes to attracting. Retaining. And also, nurturing, the next generation. Of excellent. Talent. To work. For european, institutions, but i completely. Agree with stefan's, comment, it's not just even working. In europe, it's actually working for european, institutions. Because. The big, asian, and north american. Companies. Which dominate. Ai right now, they all have excellent. Research, labs in europe so you don't have to leave europe, but you're not working for a european, institution. Elise, as martin mentioned, is a. Grassroots. Initiative. And is trying to address these challenges. With a three pillar, strategy. Which, the main goal is to. Contribute, to achieving. What we all agree. Upon, i think which is a european, technological. Sovereignty. In ai. So what are the three pillars. And they're very aligned with what has been mentioned so we focus a lot on on basically, talent, and infrastructures. So, we want to invest, in existing, talent, which we call, fellows. Which will be. Excellent. Researchers, in, machine learning and related, areas. We also want to foster. New talent. And that would be through the students. And then, as it's been mentioned, we want to build a network, of outstanding. European. Ecosystems. Which we call sites. So i'll just very quickly explain the three pillars and then go with my questions, so, uh on the, um. Fellows, side of things, we established. A year, and a few months ago. A wide range of, fellow research, programs, which are devoted, to, all sorts of topics from more theoretical. Machine learning to applications. Of machine learning, to health, or climate, sciences, you know or human centric elements, of ai. The second pillar is basically, about attracting. And inspiring. And connecting. The best phd, students, in europe. Through the ellis phd, program. And the last pillar. Consists, of the creation. Of a network. Of ellis sites, that are located. At leading. Existing, institutions. Or we have a couple examples, of maybe. Sites creating, from scratch. And all of them perform, competitive, modern ai. Across europe, and, israel. With. Economic. And social, positive, impact. Analyst, unit. Has to commit to invest at least 1.5. Million euro per year, for at least five years. Requires, a commitment, to excellence. Through international. Peer review, and also a contribution. To the network.

Towards Supporting, network activities, like hosting. Exchange, students, or faculty. Leading. And creating, research, programs, organizing, workshops. Etc. Beyond this scientific, excellence, and i wanted to mention this because it has been brought up in the discussion. The lsu, nets, encourage, its members, to have fluid relationships, with industry. With the, european, industry, to spin off companies. We also. Ask every unit to develop, a diversity. Strategy, which hasn't been mentioned, yet, but is one of the big challenges. And also to carry, outreach. And education, actions, because it's fundamental. To build trust. In ai, and also to attract the new generations, to this field we have a huge lack of young talent. The official, launch of the first 30 alice units, took place on september, 15th. And, they include, most of the, best. Um. Research centers in europe in ai, including, four right now from the uk, in cambridge. Oxford. Ucl. And edinboro. And the overall commitment, is above. 50 million, euro of their own funding, per year, for at least, five years. And this is what we've done, but, and i think this is great progress, but i think we really, i think needs need more needs to be done, and i think it needs to be done now and that's why, we um, also, drafted. A call for action, and this is a little bit, where my question, is going. Um. We believe. That, um. Europe has an opportunity, to contribute, to the ai field. With. A european, sort of like flavor. We believe that we need, for our own survival. To be producers. Inventors. Creators. Of ai, rather than just consumers. Of the ai, that has been created, by others, namely by north america, and asia. And we firmly believe, that the kind of ai that we can create in europe, will be different. From what is being created right now in north america in, asia. Probably, more aligned, with our values. And really focused, on ensuring, progress. And to me progress, means the improvement, of the quality of life, of people. Of all people, not just some people. And also the rest of biological, beings and the planet, itself. Um, so my question, is you know we have this. Initiative, that martin has highlighted. As an initiative, having great potential, you know ellis. So, i would love to hear, your vision. Your opinion, on, how do we take the next steps, in les. And how do we get. The governments. And pan-european. Institutions. To work together. In achieving this vision, what do you think would be the most effective. And the most efficient, way. To be able to realize, this ellis vision. Given, at the same time the sense of urgency. That we have, and also given the diversity. That characterizes. Europe. But that's my question. Thank you very much uh luria for those, remarks, and the questions. Uh, we have, one other question, from, uh one of the audience, and that is that ai, will automate. A lot of jobs. And the question is. You know, what can be done. Eventually, there will be new jobs but there will be a long transition, period, where people need to be retrained. And the question is what can. We do. In that respect. And a second related, question is, that we're focused, on, ai, talent, but this is mostly at the university. And postgraduate. Level. But there's quite a lot of. Use of ai, that can be done at the technical, level.

And So. What can we do about technical, education. So that sort of summarizes. The questions, from. Uh, stefan, and nuria. And. And from the. Ben upton. Uh, from the audience. And perhaps, uh martin, could you go first. Yeah i can do so, try to be brief, first of all my view is to transform, alice into what i would call an, intergovernmental. Agency. Similar, like the embl, structure. The advantages. You mentioned the salary, you know that ember, salaries, are much much higher than the sellers usually can pay in individual countries. Due to european, standards they paid european salaries. They have no taxes. Taxes to pay for et cetera. So salaries are pretty high and amber, and that makes it quite competitive. Um second i think the idea to have competition, with industry is not that new when i was young. My. In my field the the the. Best competition, was was, he was in bellaps, in the u.s balance was a prime place to go, being paid for by industry, and we had to compete with this uh, and, third. We have to make sure that our own institutions. Are competitive. So we i have to ask ourselves. I mean me and max blanc and anybody else, what can we offer to be competitive. In max blanc the, the answer is our answer is pretty simple, we offer our leading scientists, a word. A. Lifelong. Independent, research without, having the need to to, ask for funding anymore for proposals, they can do whatever they want to do, for 20 30 years. Um, having good infrastructures, and i can say that is unsurpassed. Even in industry you don't get 30 years, of free research, on your own whatever you're interested in to do, so i think we all have to ask ourselves what is our, what is our um. What is our, benefit, in. Respect to others new jobs. I feel, besides, alice we need the second initiative, which is much more, aiming. For industry. Aiming for let me say small business, aiming for those people who need to to use ai but not to invite event ai. As new i think alice is a basic, science initiative. But of course, the broad. Industry, will also, benefit, a lot if you would know how to use, ai, um which does exist and i think that is a totally different initiative, which would. Be pretty developed in parallel to. The basic, science initiative, alice. Um, antoine, do you have any, remarks. Yes quickly because i i suppose that we are out of time and uh and. Of course, we could, need to a couple of hours, to to comment, and answer, to all the, stephan, and maria, said, so so very briefly. Fundamental, research, i'm absolutely, convinced, that we need as stefan said a lot of fundamental, research. In particular, in mathematics. As he said. And probably. It can be, a, differentiating. Element, for europe. And i think it's also related, to. To elise. How how will you will we win, uh. The sovereignty. Question. Uh or address the sovereignty, question. I think that, the key point is that. For me all the all components. Of the europe, society. Must make the choice, of ai, it means that science, has to make the choice, but industry. And society, in the large, and also decision, makers. And i'm not sure that it is the case today. And it's clear that if you're a scientist, a right scientist. A smart scientist. In ai, and if you have the choice, to, collaborate.

Ways Huawei. Or or google, or facebook. Or, nobody. Then, as complicated. To make the choice of nobody, so it really means that we have also to develop. Ai, in european. Companies. In order. For them to propose. Collaborations. To to scientists. And that's for me, a key point. And of course these questions, of, of fiability. Basis. Mentioned, by stefan, are all absolutely, crucial. If we want to have an ai, based, on our values, on our history. And an ai. Accepted. By, by people. Concerning, briefly a brief comment concerning, the elite, initiative, i think that ellis, is a fantastic. Bottom-up. Initiative, as, it has been said. But i think now, it's, i think time to make some. Kind of top-down, i mean by that that this. Is easily set. As in some sense. To be. I don't know supported. By some institution. I think. I fully agree with martin. And, it's now time to go to a second step. And it means that also the the father, and the mothers, of allies, have to accept, that their child. Is now, not so big but. Big. Big enough to have it to have its own life, and to accept, that. Other people, take care of, it. Because it can be only at an institutional. Level. And i think that the, example, of embl. Is a is an interesting, example. Perhaps it has to be adapted because, ai and biology, are not exactly, the same and they don't. Exactly, behave the same way but clearly we have to do something. Which is. Which is, clearly. Needed today. And concerning, the talents. I said at the very beginning that i'm convinced, that we need. Talents, at every level i mean technicians. Undergraduate. Graduate. And this means also why we have to work with a university. Because it's really a question of education. And education. From my point of view is important. For all the specialists. At different, level, but also, education. Is important, for the next, generation. Of decision, makers. And also, for next, generation, of everybody. And and once again if we want to take. If we, if we want to have an ai which is beneficial. To, everybody. And not accept, knowledge, of a limited, number of people. Then education. Is from my point of view a key point. Thank you. Thank you very much antoine. I will be very brief in my summing up it's very clear that the questions. Uh are not easy. Uh they have no easy solutions. Uh antoine, pointed, out the. Balance, we have to have between. Regulation. And operability. And utility. That if you have too much regulation, and you get in the way of, progress, and. Innovation. That is also not helpful. Martin, pointed out one very interesting, thing which is that we must try to incorporate. Ai. Into our strengths. If we want to scale up, for example, europe, has. Large-scale. You know, manufacturing. Pharmaceutical. Industry. Medicine. Etc. And, it is only by integrating. Ai. Into, existing. Industries, that we will, actually be able to, develop, the kind of scale. That will allow us to be. Competitive. And finally i should say. It is of course very important to be ethical. Have respect for privacy, etc. But that alone, will not win. The race. You can see billions, of people signing, up to. Companies. You know where there are very few safeguards. And. You know, so i think. We have to somehow, compete, with the best, product. And the most useful, product. And that, and finally i should say. We cannot do this on a country by country basis. Because we don't have the scale. That the us, and china have. And so it is a very strong. Argument. For having. Europe-wide. Standards. Europe-wide. Data, interoperability. And europe-wide. Collaboration. So i think i'll. Close there. And i want to. Thank my co-panelists. Martin stratman, and antoine, petit. Thank you very much for. Getting together. For this, event. And i also want to thank. Uh the other contributors. Uh martial, fiat klovska. Stefan, malat, and, luria, oliver. Thank you very much. Thank you thank you sina pleasure. Thank you. Ciao. You.

2020-10-13

Show video