Use of Artificial Intelligence in the Defence Sector by Dirk Hoke - Webit.Festival Europe 2019

Yeah, of course, it's always a difficult topic, especially at a digital conference, to talk about defense. But I think it's also important to talk about reality: what's happening in the world, and of course also how we protect our wealth and our social systems. When we talk about artificial intelligence and defense, the first thing people say to me is about killer robots and killer drones. That's the first association people have when you combine the two topics. I think that's definitely not the key subject. First of all, we live in a world that's very volatile, as you just heard from Aid, and it's constantly changing. Unfortunately, there are not only good people living on this planet, so we have to deal with these kinds of things.

The second thing is also to understand what artificial intelligence means in combination with defense. In our company, for example, artificial intelligence is mostly used for totally different purposes, and mostly it's machine learning. We do, for example, satellite imaging. You use your smartphone – half of the satellite pictures that you have on your smartphone, whenever you use any kind of service, come from our company. What we also do, in order to provide these services, because you have a lot of clouds on pictures, we select pictures without clouds. For that we use machine learning, and that's quite a complicated process because you have to teach them the difference between snow on a mountain and clouds. That's quite a difficult process for a computer. So we are very advanced on machine learning. We use it in all aspects of our life: in developing our products and of course also the execution of our services.

The biggest problem that we still have in the discussion on defense is the differentiation between artificial intelligence, machine learning, and the topic of autonomy. When people come up with the topic of killer drones, killer robots, they always refer to the topic of autonomy, which means there's a machine that becomes more intelligent, that thinks for itself, and can take actions and decisions for itself. This is definitely not happening in Europe. It is definitely not happening in companies like ours, where we use artificial intelligence of course to improve our products and systems, but we have a very clear charter that for all the products we're going to develop, there's a man in the loop. So it's a very big difference: thinking about using artificial intelligence for the benefit of humans versus going towards autonomy. Still, I think it's important that we understand, as Aid has already explained, the tradition of warfare has changed. It's not a warfare of land, marine, air systems anymore. It's very much a warfare of cybersecurity, and very soon also on space.

We have also to see that many companies and countries have started to develop products and systems for space warfare. It's also very difficult to differentiate spy satellites versus satellites with robot arms and cameras that can be used for different purposes. So there will be a military aspect to everything that is done in space coming up. As you have heard, President Trump already clearly expressed his ambition to create space warfare, and we also heard the Minister of Defence in France saying that the definition of the strategy of defence for France will include from now on also the topic of space. So it is something that will be in our life whether we want it or not. Artificial intelligence is playing a major role because what we can see is that development on the IT side is much faster than on the hardware side, so it is going to influence all our products.

This digital benefit means that in the future we have the connections, of course, for military satellites. We have different kinds of air systems, marine and land systems, and it's all about sharing information, having the intelligence to understand what's going on, to share and analyze the information in real time, and distribute it back. This is something that we're going to realize in what we call the Future Combat Air System, and this is really important for Europe's development. Here we talk about the next generation of a future combat air system, which is supposed to be much more than just an aircraft. It's a combination of different systems, what we call a system of systems. We will combine land, marine, air, space into one combat system, enabling standardization of communication, standardization of connectivity, enabling to share the knowledge of all these situations, and ensuring that we distribute back to the systems only the information needed in order to be in control of the situation.

This is already happening today. We have tested, for example, a few months ago over the Baltic Sea, to fly with a Learjet and independent drones to fly information and swarms, and then have independent reactions to situations where the drones have to decide for themselves and react, and then come back to formation. So this is really working. We have tested that already, and many other nations are working on similar systems. So the intelligence going into the systems is advancing at lightspeed. Same as you see in your private environment, this is also now taking influence in the development of these new systems. What is important is, of course, also to be able to secure that communication is not changed. It's not only about interruption of communication. In the future, it will be very important to understand: is the information received still the information that was sent? Was the information modified on the way? Because as you see in many software developments, and even on smartphones, the progress is so fast that it's literally almost impossible to differentiate if a video or an audio was modified or not. So you don't know if the original information is still the same information that you see or hear in your current situation. So it will be very important also to guarantee that information is the same as sent and received. This will be very crucial for these kinds of systems, because if we cannot ensure that we have unsupervised communication between all these systems, yet we have secured that cyber security is the highest standard level going down to chip level, we cannot secure our prosperity in the future. We will not be able to handle these kinds of systems.

So machine learning will be in every part of the development of these products and systems. Artificial intelligence – first of all, I'm a strong believer that it's still brute-force computing, but on the way towards artificial intelligence, all the technology leapfrogs we will see will go into these systems. What is different from the past? In the past, we developed systems for 20, 30 years, and when we finished and went into serial production, it was based on technology 20-30 years old. This is starting now, and we'll go into initial serial operation early, 2040, 2045. But imagine in 2045 running a system which is based on technology of 2018-2019 – not possible. So what will be the challenge to all the companies and countries involved is to leapfrog technology several times along the development. That means before we even deliver the first serial production, we have ready to release two or three upgrades to be into operation with a state-of-the-art system. That has never been done in Europe before, and this is something which will be a tremendous challenge for all the countries and companies involved. But it's also a unique chance, because I just had a discussion with the president about involvement of smaller countries into a future combat air system like that. This is exactly a great opportunity to demonstrate that we are really working together in Europe. Here we need the knowledge and capabilities of all the nations in Europe to combine and to avoid fragmenting systems again. So here we can work together on research and development. In order to develop this system, we have to cross technology barriers that we have seen in the past, and we have to develop new systems. But it also gives smaller countries the opportunity to participate and to join and add to these kinds of systems.

What we also do is we start using the sensors in our products which we have not in the past. So in the future, a product like an A400M transport aircraft will have up to 250,000 sensors, and all the data will be collected in order to improve predictive and preventive maintenance. It will be realized, analyzed in real time, consolidated in data lakes, and analyzed in order to support a customer, to reduce maintenance, and what is more important, aircraft on-ground time. So this is currently in development. What is, of course, the requirement of our customers is that all kinds of data like that will be stored in a private cloud. So this will not go into Google and Amazon; this will be stored in national private clouds in order to ensure security at the highest level. But it gives the nations a totally different operability and availability of their systems in order to be more efficient than they have been in the past. So we will see AI and machine learning will be a very dominant part of the further development of products and systems in defense, especially also in Europe. Unfortunately, our world is not getting better. It is more volatile, more complex. We see more crises coming up. So we also have to understand in Europe how other countries will develop systems, and we have to understand it in order to prepare to defend ourselves. So artificial intelligence will play a major role. Autonomy is, for us, a question at least for European companies working in defense. We are committed to work up to the level of ethics that we have defined in our home nations, and it means that the man in the loop will be a fundamental criteria of further developments in the future. Thank you very much for your attention.