Bernard Ourghanlian, guest speaker at EPITA’s 2020 Research and Innovation Week
September 14-18, 2020, EPITA organized the latest instalment of its Research and Innovation Week for 2nd year students on its campuses in Paris and the regions. Packed with workshops and talks with insights on research issues and challenges, the event offered tomorrow’s engineers the chance to meet a wealth of cryptography, artificial intelligence and medical imaging experts, including EPITA professors and researchers, several alumni and distinguished guests from the world of new technologies.
EPITA Research and Innovation Week is traditionally opened by a prestigious guest speaker. Preceded by Cédric O, Secretary of State for the Digital Economy and Philippe Bournhonesque, Chief Technical Officer (CTO) at IBM France, this year the engineering school welcomed Bernard Ourghanlian, Chief Technology and Security Officer at Microsoft France, to launch the 2020 event. His speech covered a wide range of fascinating subjects, from the role of ethics in research, to the prospects opened up by quantum computing. EPITA took the opportunity to interview this consummate professional, who has a ‘thirst for knowledge’ and for sharing it.
Bernard Ourghanlian: The first reason is that I began my career as a university teacher and researcher. I taught in secondary schools, colleges and university institutes of technology (IUT), and on two-year Master’s programs at different universities, including Centrale and at ESME Sudria, an IONIS Group school. I taught for a very long part of my career. It was of course a full-time job at first, but gradually it became something I did alongside other work. Not for the money, but to maintain contact with the students. The fact is I really enjoy teaching. And to be honest, the only thing that bores me in this field is the same thing that eventually led me to also rapidly move my career elsewhere: repetition. For me, as a mathematician by training, I find teaching is too often repetitive. Having to repeat the same thing two years on the trot is something I can’t stand. At engineering schools, for example, most mathematics I taught was invented before the 20th century – well, I’m exaggerating a little, but this isn’t far from the truth. This condemns you to repeat yourself and it can inevitably get boring. It was for me. But I do still very much enjoy interacting with students. It allows me to understand their aspirations and what they’re interested in, and how they prepare themselves for the world of tomorrow. It also sometimes means I can pass on a few messages about certain subjects close to my heart. So, I was absolutely delighted to be able to address the students at EPITA during this conference.
Yes. You could say I have a thirst for knowledge. I’m interested in everything, including non-scientific fields such as literature, philosophy and sociology. I’m also totally crazy about poetry and have always been a musician. I’m attracted to everything that expands my knowledge. I need to learn. This repetition factor is crucial in who I work for. I would never work for an employer where I could not constantly be learning. That’s why I’ve stayed at Microsoft for all these years. There’s always something new to work on and innovation is practically limitless.
Quantum computing: a key research focus
You have to be curious, that’s for sure. You must also have the desire to learn every day. If you’re not learning anything, it’s a day wasted. So that must be your approach. But not your only approach: you also have to be open to others. It is rare for innovation to take place in isolation, in a bubble – there are of course several counter examples, but these are extremely rare. You need to be capable of working with others. Curiosity – the desire to learn, to discover new things and discover what we don’t know. This is the driving force of innovation. It’s also an extremely powerful driving force on its own.
Yes, this is one reason why we innovate. Behind innovation there is research. And behind research there is this idea that there’s a sort of dichotomy between fundamental research and applied research. Personally, I don’t like this dichotomy because it is inherently tied to questions of temporality. For example, Microsoft’s work on the quantum computer dates back to 1997, so I wasn’t there when Alexei Kitaev wrote his first paper on the topological quantum computer. When we invent, what really matters is that the invention can be used. And when that day comes, it becomes an innovation: an innovation is an invention that finds a usage. For as long as it’s not being used, for as long as uses have not been found, it remains an invention and not yet an innovation. There are also a lot of issues around usage. Taking COVID-19 as a contemporary example: the pandemic has, paradoxically, made us learn a lot about the world and about ourselves. On the one hand, we’ve asked ourselves a whole host of questions about the meaning of our lives and about more philosophical issues. It has also made us realize that working from home, working remotely, wasn’t straightforward, for sociological reasons. Typically, in various companies, remote working is still difficult because it calls into question certain managerial practices, particularly those based on the idea that the manager is there to check that people come in in the morning and leave in the evening, that they’ve done their hours. From a ‘clocking in’ perspective, this conveys the idea that we need to show up, to arrive before the boss, and leave after them. This demonstrates something important. It shows that the evolution of management has not yet taken place. Beyond that, we’ve also realized that working from home isn’t always straightforward because we don’t necessarily have the equipment and because it was complicated by childcare. When there are two of you working remotely, the distribution of roles within the home, within the family, comes up. Even when this question had already been settled beforehand more or less equitably. The situation has made us revisit all this. And behind this more or less forced remote working, so many Microsoft researchers, anthropologists and sociologists, have studied the significance of being confined to our homes and working remotely, and realized that many things needed to be improved. This has raised questions about people’s set-up at home – some people were working from their living room table and, had severe back pain after one hour – and about relationships between couples, the parent-child relationship, the search for meaning. Because of it, so many of the questions that came up went beyond pure technology.
During his talk, Bernard Ourghanlian discussed several projects at Microsoft laboratories, including the Emma project
As part of Impact AI, we seek to ask questions about the impacts and uses of AI, its ethical issues and brakes, about innovative topics. Our strategy is one where doing things in isolation makes no sense. AI mobilizes the collective unconscious (ever since 2001, A Space Odyssey and Terminator) and has been fed by a certain form of fear. If we want this field to move beyond mere invention and become truly useful for as many citizens as possible, we must go out and meet them and not just stay in the corner. But it’s not as simple as that. Here is a typical example of the failure of innovation. Apple was seen as having introduced a major innovation in the iPhone and the iPad, with their touchscreens. But Microsoft had introduced a Tablet PC back in 2001. At the time, we thought that by using a stylus, the possibility of using writing (even for those who write very badly) would be a new way for humans to interact with machines. But we were wrong because we missed the point: instead of imagining an interface we could interact with using our fingers, we thought people would continue to write with a stylus. It wouldn’t have been very complicated, but we hadn’t imagined that concept! Specifically, we were wrong about the usage scenarios, even though we had brought out this computer almost five years before Apple’s touchscreens. This shows how important it is to understand how a technology will be used before introducing it to the market. Because when we invent, what matters is how the invention will be used. An innovation is an invention that finds a usage.
Artificial intelligence. This is no great surprise, as I became familiar with AI during my studies, but in a very early incarnation based on rules. AI was an outmoded subject and, in this sense, computer science is a succession of fashions. Over time it has alternated between centralization and decentralization: mainframe computers gave way to PCs, which made decentralization possible, before the Web recentralized it all. We see that computing is not immune to fashions. AI has been a pleasant surprise, in some ways. The idea of using statistics and having a probabilistic approach to modelling is something I didn’t necessarily expect. The joint progress made by algorithmic evolutions, data collection and computation capacities have contributed to the revival of AI, which was hampered by concepts that boxed it in. It has been a real revolution.
I really think it is. It uses the infinitesimal properties of matter, in particular a quantum concept that is very hard to understand, superposition: a particle can be in multiple states simultaneously. That may seem counter intuitive. Instead of having a switch that is simply open or closed, a potentiometer enables all possible positions, from full off to full on. The purpose of quantum computing is to be able to use this phenomenon – there are several, but this is the main one – to be able to do parallel processing. If we take the example of MIDI Maze (one of the first games played on a computer, where you escape from a maze), the conventional computer will program a path to see if it works and try the others in succession. It is possible to do the processing in parallel with several other computers so that they can explore all the paths. This will take a long time. With a quantum computer, you can do this work of exploring all paths at the same time in an exponentially shorter time. We can therefore imagine replacing certain algorithms with exponentially shorter quantum algorithms. Quantum was popularized by Shor’s algorithm, an algorithm for integer factorization. With very big numbers, this could take a very long time. This technology is used in encryption today: to break these encryption keys, a conventional computer would need a billion years. With an unsophisticated quantum computer, it would take about one hundred seconds! If we can have a large enough quantum computer, capable of scaling up, this will transform the world of computing that we experience. And not just a little bit! We would need to switch to another encryption device, because we could decode all messages exchanged through the internet. But there is also the possibility of exploring certain types of processes, such as chemical processes, and the ability to create a catalyst to capture CO². If we were able to do this, we could solve global warming. This is much more than just a working hypothesis and it could be the answer we are looking for. But if we don’t succeed, we’re facing a terrifying world. The quantum computer gives us access to this kind of dream.
Bernard Ourghanlian joined Microsoft France in 1999 as Technical Director. He was appointed Chief Technology and Security Officer in November 2002 and interim Director of Services. Bernard Ourghanlian is responsible for overall leadership of Microsoft’s technology and security strategy in France. He is also in charge of technology risk management at Microsoft France. Bernard Ourghanlian has been a member of Microsoft France’s executive committee since early 2009. He also serves on the board of Syntec Numérique and Paris-Sud University. Liaising closely with senior management and the relevant bodies, he is Microsoft France’s technology representative via à vis key government ministers and major government agencies, IT departments, standardization committees and the research and education sectors. He also works closely with Microsoft Corporation, Microsoft Research and its various product groups, and actively participates in the development of Microsoft products and solutions. He is a member of the executive committee of the INRIA and Microsoft Research joint research center.
Before joining Microsoft, he was Technical Director of Digital France, where he was actively involved in designing and developing Alpha microprocessor architecture and its various support system software. Before that, he contributed to the design and implementation of the first nuclear magnetic resonance imaging techniques at Thomson. He began his career as a teacher and researcher at Orsay University. He also holds a doctorate in mathematics and has written several specialist books on statistics and computing, including a reference book on Alpha microprocessors.