Guillaume Faury CEO Airbus at Adopt AI - Special interview: AI in Aerospace Manufacturing

Yeah, good morning everyone. Happy to be here and speak about AI. Indeed at Airbus, we've been using AI for decades. The traditional AI, machine learning, deep learning in different fields that was more for experts, quite sophisticated. And things have changed tremendously in the past years with the arrival of GenAI, democratizing AI. And we've trained a lot of people, between 70,000 to 80,000 people trained on AI. We want to embrace AI, to use AI as much as we can. GenAI has an incredible potential. We've started with a number of rather small projects, proof of concept, pilot projects, and we came to the conclusion that those hundreds of projects – I think we have something like 600 projects that were launched – were not very efficient. That was probably a necessary pain, helping AI to be endorsed, to be understood better, to reduce the level of fear with AI. But it was not very efficient. So we've reduced to roughly 20 projects that can significantly change the efficiency of what we're doing, and we're focusing on those projects. So that's the application of GenAI in the company. And we see coming agentic AI with a huge potential, but we're not yet there, and we're just starting to understand how we will make use of agentic AI.

Yeah. So at Airbus we are mostly using AI in three fields: development, engineering, production, and services. Production is indeed one of our main challenges at the moment. AI is used already in automation – I mean you use robots that are powered by AI. We are using AI for quality checks. We are using AI for efficiency of the processes, and that's being either in place or deployed. But the large development and deployment of AI for manufacturing will be on the next generation of planes. So we're currently working on identifying the best way to use AI within robotics and automation, and within engineering simulation, to do better planes and produce them in a much faster, efficient way, and with an even higher degree of quality first time. So that's the kind of application we're having in AI. And in services we have plenty of applications. One that is quite amazing is the so-called predictive maintenance, where based on the huge volume of data that we have on board airplanes, we can use AI on patterns of failure. So you can identify before it happens what is going to happen in terms of deficiencies in failure, and you can prevent failure rather than fix failures afterwards.

So, we already have drones and autonomous flights or planes mostly in defense. The use of AI in defense doesn't respond to the same expectations than what we do for commercial applications. In commercial airplanes, we're today using AI in non-safety critical functions, even to enhance safety, but not for the core safety systems. Because AI today is not considered predictable enough. So we can put AI in the onboard systems. That's probably for later. But we're using AI around. So all what we do on the ground: engineering, production, quality, support and services. But when it comes to putting AI in the planes, we do it a lot in military planes. For instance, we do automatic refueling based on AI. That's much more efficient than based on manual steering. But we're not at the point where the autopilot, the fly-by-wire of a commercial airplane would be powered by AI. This is for later. In the application of AI onboard commercial airplane, what we do is enhance safety. So on top of what we have today, for instance in automatic landing systems, we are using AI on image recognition to increase the level of safety, to be able to identify, to detect objects on the ground that the pilot would not identify alone. So this is the kind of development we see arriving in aviation. We take by far more risks when it comes to defense space when there's no human being being involved, but when we are carrying passengers, then we take less risks and it takes more time to deploy the AI systems.

AI, I mean AI is plenty of different things. So that's the first remark. The second one is it gives the potential to assist humans for tasks where complexity is involved. And actually when you look at the development of the battlefield today, it is a digital battlefield. Therefore, the amount of data that has to be managed is no longer compatible with what a human being can do. So, we're deploying AI in military systems to digest a huge volume of data and bring consolidated information or help for decision to human beings. But we keep human beings at the center when it comes to decision making. That's really what we see happening in defense.

Well, when it comes to defense, that very much boils down to the concept of utilization that the armed forces will want for the systems they're using. And there's a fast development. There are different philosophies, but basically what we see happening is really keeping the human being at the center when it comes to making strategic decisions of striking or not striking, of acting or not acting. And therefore AI is here to simplify complexity and bring to the human beings information or decisions they can swallow, they can manage. When it comes to commercial, that's very different. In the commercial arena, we use AI to be able to develop better planes, to be able to produce faster, cheaper, and with a higher degree of quality, and as I said earlier, to support aircraft in service by increasing the level of operational readiness by preventive maintenance, by avoiding failures. That's basically the way we look at it.

So space is a sector of very fast development. It's a way to use data and to make data available everywhere. So the fast evolution of technology driven by digital as a space component. We've seen giants in the US going very fast, and we need to create players in Europe at scale. When we have European players at scale, like what we did with Airbus 30 years ago, or what we've done 20 years ago with MBDA on the missiles, when we create players at scale we can compete with the Americans or the Chinese. And that's what we've decided to do with our friends of Thales and Leonardo to create a satellite manufacturer and operator that will have the scale to compete with the bigger players around the world, to be able to fuel innovation, to do the right level of investments in Europe for Europe and also for the rest of the world. So we'll be a competitive player in space moving forward.

To protect Europe. It starts with protecting from space and from the air. You protect from the top. And the future systems will be like in the consumer goods environment, it will be about connecting objects and exchanging data. So it comes with a cloud, and we have to create a European connected system, which European cloud we call it the combat cloud. And the purpose, one of the purposes of the future combat system that is a German French Spanish program, is to create that combat cloud that will be the place where data will be exchanged between satellites, tankers, fighter jets, helicopters, and also objects on the ground. That's a key enabler of the future digital battlefield that is going to take place when there is confrontation in the future.

So it's not something digital, it comes by waves. We are already today connecting objects digitally. There are protocols to exchange data between different objects. But those protocols are quite limited and it's not like a network where everyone can connect like a cloud. So it will take a decade, maybe a decade and a half, before we are at the level we want to be, but there are incremental ways of connecting objects that are being applied already.

Yeah. So there's a big debate at the moment on the way Europe is approaching regulation, not just for AI but for plenty of other topics. And indeed we are in a period of fast change, plenty of new technologies. And faced with new technologies, you want to do two things. You want to protect against failure and risks, and you want as well to unleash potential to catch opportunities. And there is an offset between the two. There is a challenge between the two. What we see happening in the US is unleashing the full potential of technology and accepting the fact that there are risks and managing risks on the way. Europe is taking a very different approach, which is to regulate upfront to try to reduce or prevent risk from happening. Well, that is something we can understand, but at the same time, it's limiting the opportunities. It's reducing the speed of action and the field of action. I think today there is a need for Europe, and that's what we're hearing, to rebalance to be able to accept more risk to have much more opportunities. We are at a time where speed matters, and if you protect too much, well you just lag behind the others. And that's really the risk I see at the moment in Europe. Therefore this discussion on simplification, which in my view is not about simplification, it's about accepting to take a bit more risk to catch all the opportunities and not lose the race compared to other blocks, and especially the US and China.

I think I like to summarize with two things. We need scale. And when Europe comes together, we have the scale and we can compete and be efficient. When we are too fragmented, when we don't manage to cooperate, to play as one block, then we are slower than the others. So really cooperation, thinking at scale. And the second one is speed. And back to my previous message, I'm amazed with what we see happening in some parts of the US. I'm traveling to China quite often and China goes very fast as well. We should not underestimate the speed. And therefore this debate on simplification is in my view a debate on speed. How do we give to Europe more speed when it comes to the regulatory framework? But that's true as well for companies, for labs, for research centers. We need to be obsessed by speed because competition today on technologies is about being faster than the others. So I would say scale and speed are in my view the two priorities we need to focus on.