How Airbus Decides What AI Should and Shouldn’t Do
In this episode of the HR Leaders Podcast, we sit down with Vincent Dupuis, Vice President HR Digital & AI at Airbus, to unpack how organizations should decide what to automate, what to augment, and what must be protected as AI reshapes work at scale.
Vincent explains why augmentation, not replacement, is the real story of AI at work, using powerful analogies to show how AI should extend human capability, not hollow it out. He breaks down how Airbus thinks about freeing people from low value tasks, while deliberately protecting deep expertise, critical thinking, and safety critical knowledge.
Most importantly, he shares why ethical governance, human in the loop learning, and robust knowledge roots are non negotiable in environments where quality, trust, and safety define success.
🎓 In this episode, Vincent discusses:
How automation should free time for higher value human work
Why augmentation beats replacement as the dominant AI model
How Airbus embeds ethical AI governance before access is granted
Why protecting deep expertise and critical thinking is essential for safety
How Airbus decides which work should be augmented, automated, or protected
What if global compliance took 5 seconds?
It does now.
atlas copilot partnered with our friends at Deel, the all-in-one Global People Platform, to bring on‑demand employment matters and compliance intelligence into your daily flow. With Atlas' GenAI agentic platform and Deel’s vetted sources, get authoritative answers across 150+ countries.
Ask anything:
“Is a 3‑month probation period compliant in Mexico?”
“What termination notice is required in Germany for 18 months’ tenure?”
“How long are the maternity and paternity leave in the UAE?”
You get clarity, that used to take hours, in seconds.
This is not a feature. It’s a new operating model. The chaos of Slack threads, scattered PDFs and expensive calls collapses into one trusted interface. Your team moves faster, escalations drop, decisions align and you see regulatory change as it happens.
If you’re still stitching notes from outside counsel, you’re already behind. AI-Smart HR leaders are shifting to always‑on intelligence. Don’t let compliance feel like navigating a minefield.
00:13
Vincent, welcome to the show. How are you doing? Good, thank you. Fantastic place, fantastic day. So thank you for welcoming me. What brings you here today?
00:21
Hi, we are having reflections at the moment at Airbus about preparing the future of work, and it's a great place to get inspired, to know what the Airbus do, to open the world of possibilities. Yeah, and you're also going to be speaking on one of the sessions, right? Yeah, tomorrow I will be on a keynote speaker, and I got the pleasure to attend as well a roundtable discussion this morning on the future of work.
00:48
Yeah, but for your own session, what are you going to be talking about? Tomorrow, it will be on how do we go from megatrends from the world of the future to how to translate that into technologies and roadmap to say, OK, from the why, the what, and how. Wow. Trying to. I was about to say, that's a tough one.
01:07
It's a tough one, but I think it's where the people get stuck. It's how to canalize all these ideas and possibilities into something to have a methodology towards it. So I will share my humble approach on it, and then let's see. Yeah, well, let's get a little bit into that now. How is Airbus defining which role should be automated versus augmented or protected as AI shapes the future of work? Big challenge on everyone's mind. Yeah.
01:34
Let's say I will start by the most obvious one. It is the augmented. I think it's what you hear everyone. So augmented, everybody will be augmented when we think about people will be replaced by people using AI. Yes. Augmented by AI. So people will not be replaced by AI directly, or let's say not everyone. The majority will be replaced by people using AI, augmented by AI.
01:58
And for me, augmentation, it's like, I don't know if you know exoskeleton? Yeah. So it's like you get an exoskeleton and it enables you to climb higher than what you usually do. So it does not change what you do. You climb mountains, you walk, etc. But it enables you to go higher. It free-exes your performance. Exactly. So I think it's an image I try to keep because then, in terms of change,
02:22
It's not about changing what you do. You need to upskill to get used to that skeleton so that you can make the best use of it. So this is what I mean by augmentation. So in terms of accompaniments, it's how do we upskill the people to use these AI tools to augment their work.
02:40
I love that analogy, by the way. It's a great analogy. For automation, I think it's to get rid of low-value added tasks. And here, the analogy is like, I don't know if you have a robot at home doing the vacuum cleaner. I do have one at home. Okay, you have one. And it's just to free up your time that is for you more valuable to be your father, to be your husband, to be whatever you want.
03:03
Losing time for that is something you would rather delegate to a robot. This is my analogy. So robot vacuum cleaner is where everything you think a vacuum cleaner could do at work is what you should automate. Where you think, okay, I would rather get rid of that. And for the transformation, which is about the last part, it's where we have to be dedicated. I will...
03:30
split in two things. At Airbus, our purpose is to build airplanes. That is what we do. And if we can then maybe lighten the support function like HR, so that we automate things, so that our HR people spend more time with the people at the frontline worker, so that they can work more
03:55
and more engaged and facilitate their work to deliver more airplanes, then we succeed. I think what we do, the transformation of the HR functions is to free up time for that so that the people get more of human touch for their operational efficiency. And then when it comes to the one that we should protect, it's an interesting discussion I got last week with one of our HR leaders. Think about, you see two trees.
04:25
One tree is a worker which is 40 years old. Another one is a young guy with a high of competencies. And both say, wow, but they are both skilled the same way. But under storm conditions, stormy conditions, you see one tree is standing up and one tree is falling down. One tree falling down is because the roots were very thin and not very deep.
04:51
because their knowledge, their critical thinking, the way they learned, has not been deep-rooted in their minds. So that, in fact, this is where at Airbus we think the risk is that we grew the top of the tree. We see that, but when things get tough, for instance, on the assembly line, you get...
05:18
a rivet which is not perfectly riveted but you lost your critical thinking because your AI goggles told you it's fine and that you don't think or you have you have lost that deep knowledge then we will produce non-conformity and maybe we'll produce non-quality aircraft yeah and this is where we
05:40
We have questions, philosophical questions. It is, how do we make sure that at the end, we want to produce safe products? Because it's what our purpose of the company is. So this is what, for us, is what we want to protect. Because it's part of our reason of being. And we don't want brittle knowledge. We want robust knowledge. And I think that is what...
06:03
is what we still need to shape. So how to ensure that robust knowledge and not brittle knowledge. I love all of your analogies. It really brings you to life. And I think I agree. It's one of my concerns I have about my daughter. And because all of the answers are surfaced.
06:21
So you really don't have to go through the critical thinking of starting with a blank piece of paper and truly understanding, right? To your point, because access to knowledge is easy. It's easy. It's right there. But do you truly understand, right? And have you had to go through the critical thinking to understand it? Like that's one of the challenges we have right now. And I also love your analogy about the roots. I think we could also apply that to our data. Yeah.
06:51
If we don't have our data, which are the roots, you know, have good data, our tree is also going to fall over as well. In terms of the data piece, what are some of the steps that you're taking to embed ethical frameworks for AI adoptions? Because that's obviously a big challenge right now for everyone. Yeah, sure.
07:17
Ethics is part of our DNA and values at Airbus, part of our sustainable goals, our way to bring ethics and to have ethics at work. So for instance, the first thing, for instance, we are a Gemini company. We are a Google company, not a Microsoft company. And then we are opening. We took the decision to open fast Gemini to all employees.
07:42
just to avoid that they use it at home and at the end we have problem of data leaks and security breach. By then to do that, for instance, and then now to subscribe to the Gemini for Workspace, you will get activated only if you go through a webinar with a quiz where inside of it, it's a lot about ethical use of AI. So we make sure that the people get the first
08:10
vanish of what does it mean to take risks and to be ethical about AI, what they can do in Gemini to take CVs of people and then to do things, also what they can't do. So they see the first layer so that to get access to the functionalities, you need to go through this
08:30
must pass about understanding this ethical risk that you go. So you have that before they can get in. If you don't attend one of the sessions and you don't pass then the quiz, you don't get activated. Interesting. Okay, so that's the first layer to make sure that the people have listened and understood what are they expected to do and then you get access to. The second is about, okay, now that we have this,
08:57
Workspace, people will create gems, will create agents. Tomorrow, we have an agent space or Google Enterprise or Gemini Enterprise. And here we are building, we have a governance. So that's any applications
09:14
In my team, I have a data team with data scientists, data officers, data architects. So we have all the data roles and you don't deliver
09:31
I can't deploy a solution, whatever it is, being a normal software or being now an agent, without having the stamp of our data governance team, where the ethical use of data, that we are compliant with the AI Act in Europe, that we have a legal person as well, so we need everything to get stamped before it gets released.
09:52
Yeah, so very, very robust. Very robust. And we are very strict on it because it's part of our reputation. And it's also part of your values. It's part of the values. And we know that it's a lot about trust. And for that, you need to take it seriously.
10:07
Yeah, yeah. Oh, I love it. The training you mentioned before they get access, how is that delivered? Like a digital course or how does that work? So that one, for example, so you have a lot of self-service course on different platform. Yeah. But that one, it's not only learning, it's delivered by a person. So it's like a life. Cohorts. Cohorts. So you apply to one hour sessions. And you do a digital course.
10:34
During the sessions you do, you have quiz to make sure that people are attentive as well, are there. And you have Q&A so people can discuss. And then you have live session then. And then you receive a questionnaire afterwards before you get activated. Yeah, love that. I just wanted to like bring that to life, like how does that work? Because there's so many ways you can do that, right? I love the fact that you made it interactive and rather than just a digital video.
10:59
Yeah, so it's not a video. It's really, we want it to be human. The more you put AI in the loop, we got that once. I think we learned from our mistakes. I think last, beginning of the year, there was some sessions about AI. And this was kind of, with a kind of avatar doing it. AI delivered by AI. And it was kind of weird. Yeah, of course. So the more we go into AI, the more you need to put human in the loop. Yeah. I think that's essential.
11:28
When we think about the capabilities of HR across the organization, what are the capabilities that you feel are essential for HR to lead this human AI transformation? Something which is not easy for HR, the capability to unlearn.
11:48
Unlearn. Unlearn. Yeah. Because on the last 10 years or maybe 20 years, I think the world of HR stayed unchanged and they were quite successful in their way to hire talents, to promote talents and to manage all the workforce. And I think we need to unlearn because everything that was true before may become obsolete. So I think it's not easy to put in a posture of, okay, I need to...
12:15
unlearn what I thought was true until now. And then another change in that unlearn thing is everything was a lot driven HR for HR. We see here, a lot is HR for HR. But you will never, I think, get benefits of the full AI if you don't think that HR is part of the workplace now. It's part of everything. To see it from...
12:43
a user standpoint. For instance, I love in Microsoft, they don't have a chief HR official. They have a chief people official. And I think that changed the perspective. It's for the people, it's not for HR. And if I take the example, if we continue to do HR for HR, I see some agents, let's take that example that, you know, very common one, if you want to raise your, a day off,
13:12
An agent will do it for you. So you don't need to connect to the system to apply for a day. You discuss with an agent. And you say, I want to be off next week. OK, do you have the credits? Fine, but that's it. I think that's really what AI should do. But if you are off next week, what I want the agents to do is say, you know you have a meeting with these people.
13:37
Should I then decline the meeting? Then should I put automatically an absence note in your calendar so that this is holistic? Who else is picking up the work now? Exactly. So I think we need to think. So if adjunct AI is not doing that, it's limited to HR for HR, I think we missed the point. And you're right. Most of the tools also are good at one thing. And to your point, it doesn't do the rest. Exactly. It can book the holiday, but what about...
14:06
They're not going to accept all of the rest. And that's why HR needs, my belief is that HR needs to open to the rest of the discipline with the workspace. Otherwise, the people will say, okay, hands.
14:20
So I think we miss opportunities. It's a moment for, yeah, exactly. We don't want to miss an opportunity, right? And then the last thing I want HR, I think the future work of HR is to be then the architects of the future of work because a lot of things will change. So to think in that moment holistically, but as well, if we think that with the new generation, that they are more individualist,
14:50
that they can work remotely, that then why should I come to work in the office? But then who is defining the why I should work at the office? What is defining what is a team? And if we think it is important, who else than HR should architect their policy, their processes, so that these elements are elements that matter. We reward talents.
15:17
If team spirit is essential for the future and not individualism, how do we reward team in future? How do you create teams in future? Winning teams, not collections of talents like in football, you know? Best teams, if you put champions together, they don't win championships. They've seen it happen, yeah. Sometimes it's because there is that dark matter that unifies these people that makes the team performing.
15:43
And then I think the HR needs to become a bit of this philosophical question, this social questions. And I think that to become the architect, they need to have this new discipline in the loop, not only HR processes. That's my belief. Yeah. Well, listen, I know you've got to go, but we need to do a whole show. So let's revisit this because there's so much more to talk about. But I really love the way you brought this to life through your analogies and your storytelling as well, which is really, I'm going to steal some of those analogies.
16:13
But enjoy the rest of the event. I'll see you soon, all right? I'll see you maybe tomorrow on the stage. See you. Thanks.
More from the HR Leaders Podcast
Vincent Dupuis, Vice President HR Digital & AI at Airbus.