Post Next: Siemens CEO Barbara Humpton on the 'fourth industrial revolution'
20:46 Share

I'm David J. Lynch, global economics correspondent for The Washington Post. Welcome to Washington Post Live. I recently spoke with Siemens USA President and CEO Barbara Humpton at a Post Live event about the future of manufacturing. We talked about how artificial intelligence could unleash what she and many others call a fourth industrial revolution, which would require new ways to train the future workforce.

As we heard you say in the intro, we're currently in the midst of this fourth industrial revolution. For folks who may not be entirely familiar with that term,

How would you define that and what's Siemens' role in that revolution? - Yeah, well, and what's so cool is that Siemens has actually been part of every industrial revolution. Siemens actually came to the United States 160 years ago. And you think about that first industrial revolution where steam was used in order to drive the manufacturing lines. And then we saw the introduction of electricity.

And then we saw the introduction of automation into those lines. And here we are in the fourth industrial revolution, and some even say the fifth, as we truly move into that era of applying AI. Now, for decades, people have thought of manufacturing as the old-fashioned sector of the economy. And in fact, manufacturing has not changed as quickly as our IT and digital sectors have changed.

look at this, now the industrial and the technology sectors are coming together. We've got software-defined automation, where now we can make changes on the factory floor at the speed of innovation in the IT sector. It changes everything. And so I really look forward to talking to you about the miracles that are happening in modern manufacturing. It's interesting, because NVIDIA's CEO, Jensen Walsh,

But Jensen Huang said recently that all American companies will eventually need or will become artificial intelligence factories.

entities that produce both goods and an AI version of those goods. It sounded to me a bit like your digital twin platform. And I wonder if you could tell us a little bit about what that is exactly and why we need digital twins. I'm so glad you're asking this question because we live in a world where up until now to create something new, we might...

First dream about it, sketch it, draw it on paper. We might then start to build prototypes, bend metal, try, try again. In the real world, that takes a long time. In the real world, sometimes we can't even recreate the conditions that we want this new thing to operate in. Think about the Mars rover.

NASA and JPL actually used a digital twin. They virtually modeled that rover that was going to land on Mars. They modeled the conditions of the reentry to figure out what's the right set of parameters that would create the perfect landing. And in fact, that rover worked and it lasted. That digital twin gives us the ability to model something we want to make,

gives us the opportunity to model something that we want to use to make it, like a factory, and it gives us the opportunity to model how something is going to operate in manufacturing, in its operations. Now, something that Jensen loves to say, and I love it every time he says it, is that Siemens is the operating model for our real world.

Because, yes, he's been building things for our digital world, but here we are in the world of atoms. And we're translating, we're combining the real and the digital, and it's transforming the everyday for everyone. Hasn't some of that capability been around for a long time, though? The ability to sort of, through software, call up a representation of something on a terminal and play with it? In what ways...

does the digital twin element of this advance that capability? - No, in fact, you're absolutely right. The digital twin has been under development for decades. Siemens made its first investment in a software company to do just this in 2007, the same year the iPhone was introduced.

And in that time, we've seen really innovative leaders stepping forward. Imagine all of the auto manufacturers. Imagine all of the semiconductor manufacturers we've been talking about this morning. The shipbuilders, et cetera, they are using digital twins because, frankly, the stakes are too high.

We have to be using these tools in order to speed up and drive efficiency. So now this is becoming more common. We're getting to that point where everyone can have access to the digital twin. The compute power isn't as expensive as it used to be. The tools are more ubiquitous. We've even seen, for instance, colleges establishing digital foundries.

teaching students how to build digital twins so they can serve the small manufacturers in a region and help those who can't afford all of the big tools that the auto manufacturers have been using. They can now get that benefit in their own operations.

Let's talk a little bit about industrial AI and the ways in which that differs from the sort of consumer facing AI that many folks in the audience may experiment with, chat GPT and the like.

What I find as an individual using effectively the free version of these AI capabilities is that there's a lot of mistakes. There's a lot of hallucinations. You can't really trust what you're given. You don't have that kind of margin for error when you use industrial AI. How do you get away from that phenomenon of hallucination and

unreliability, and how different is industrial AI from what the rest of us may be used to? - Well, yeah, that's right. First of all, AI. What is this large language model that we're now using so often in our everyday lives? It's a prediction engine. It's a prediction engine.

And you can't rely on that engine to give you the same answer every time, and you can't rely on that engine to necessarily give you the answer you're looking for. It all depends. Depends on the data that's coming in. Okay, now take that same technology, but put it into an environment like a factory.

And by the way, David, factories produce incredible amounts of data, but it's never been possible to draw conclusions from that data. There's just simply too much, too many connections, too much coming at any of us as we stand in the middle of that factory floor. What is going on here? But the models now applied in that very controlled set of data that's coming from known sources, that becomes a very reliable model

So this industrial AI is quite different. And we've been speaking to legislators, that little clip included part of my testimony earlier this year, making the case that as we regulate, we should think differently about industrial AI and the lower risk that it presents, the greater opportunity it presents to improve the efficiency and improve our operations overall. - Interesting.

Can you give us an example of how you're using that today or perhaps a way in which it's already provided you with a breakthrough in your operations? Yeah, I love these examples. Think about an auto manufacturer who says, gosh, I've got too many errors on car doors coming through my manufacturing line.

And they put out an RFP and consultant after consultant comes forward and says, you know, I'll build you a system that will find this kind of error or that kind of error. And we brought in AI. And what we said to the manufacturer is define perfect. And then what we'll do is compare every car door to perfect. Perfect.

How would that change your operations if you could detect an error as soon as it occurs, not have to look and count on chance as things come down the line? Now, another way that AI is being used in these manufacturing operations is to actually detect, prevent,

breakdowns on the line. You know that keeping manufacturing lines up and running, that's the key to strong financial performance. And so the ability to use all that incoming data and ask ourselves, is there some component on the line that isn't performing as expected? Should we speed up the maintenance action on that component? Can we repair this now before things actually break down? These are kind of the early, early proofs of the power of AI in our environment. Yeah.

- Wow, and how does the availability of this technology impact the skills that are required of your workforce? - You know, this is, I'm really loving this question right now because in the past, think about every other technology that we brought into the world. Gosh, we had to have courses to bring people up. I've heard a lot of talk about, even this morning, what will it take to reskill Americans?

This is the first technology that comes to us. How many of us are aware that we're using AI every day, right? When we pull up a map and we say, I'd like to go from my office to the Washington Post, AI is at work behind the scenes finding the optimal route, and we don't even know it. There's one more thing about AI, and that is it brings expertise to the least skilled among us.

For those who are experts, for that individual who could walk into the factory floor and just by listening to the hum could tell us if a machine was about to break down. Mere mortals don't have that skill, but the AI does. So AI is lifting the skills of those who are just entering and just learning. And then likewise, it's very approachable. So two examples.

We are down in Fort Worth, Texas. You know the AI data center revolution that's going on right now is requiring a ton of electricity. And we can't build the electrical switchgear, switchboards, et cetera, fast enough. And we needed to open up new operations.

We found an industrial park in Fort Worth, Texas. We built a digital twin of the line we wanted to put in. We built a digital twin of the products that we're going to be building. Then the question is, who's going to work here? Well, I mean, Fort Worth, Texas has a lot of manufacturing, and so manufacturers are scarce there. The team hired high school principals and teachers, brought them in, and they built a curriculum.

and we're bringing folks in off the sidelines. People are coming into our facility, they're getting some classroom training, then they're getting experience on the digital twin of the equipment they'll ultimately be exposed to, and then they're coming onto the manufacturing floor. And so we're meeting former chefs, former delivery drivers, who are now manufacturers.

and even better, a former checkout clerk at Costco who was so good during the training, they asked her to stay on and be a trainer for the next groups that are now coming through. This is an opportunity for us to use AI

a language, a translator, so that people can come in from the sidelines and now speak English to the technology they'll ultimately be working with. I can't wait to see what the Fort Worth team does next. - Have you had any former journalists show up in Fort Worth? - Funny you should ask that. I'm gonna tell you. - The way things are going. - We invited a journalist in a couple of weeks ago, and I think she came in with this expectation that first, the data center bubble is gonna burst, and what's gonna happen to this factory? And then second,

Aren't the robots coming to take all our jobs? And by the end of that visit, what she said is, well, when AI eliminates my job, maybe I want to get into manufacturing. Let's talk about that, though, because as you know, there are tons of predictions about what AI will mean for aggregate demand for humans in the workforce. And with every new wave of technology, there are always these concerns that it's going to render human workers obsolete forever.

The one question I have with AI, though, is whether the pace of change may be faster than the labor market can handle the adjustment. In other words, the growth of this intelligence just so dramatic that it may automate more and more positions faster than people can find new work. What do you think? I'll be provocative and say I hope so. Let me tell you why.

In the world of manufacturing, we have perennially 500,000 open positions. And some forecasts say we're going to need three to four million more people in manufacturing to produce the things. I mean, David, we're talking about the real world, right? It's one thing to have information flowing. It's another to actually produce the things that we need. So the real world needs AI, right?

And frankly, getting more people used to the idea that our next jobs may actually be in fields where we're the ones engaging with the real world. Once upon a time, we all thought we would be a services nation. But the fact is that if you make, you innovate.

We need to manufacture in order to be a thriving and secure country in the future. So we do need more people coming into fields where they can apply their know-how in the fields of manufacturing. Are there jobs that are going to change? Absolutely. And what we're trying to do is get the message out not to be afraid of it. From the first time a human picked up a rock and used it as a tool,

tools elevated the role of humans, and this is no different. This is a moment for us to step up and get creative. - Yeah, I wanna ask you about where young workers fit into this. We came across a study by the New York Federal Reserve recently that said labor conditions for recent college graduates have deteriorated noticeably in the past few months. The unemployment rate for recent grads is now up to a four-year high of almost 6%.

Yet manufacturers already say they can't find enough workers for these openings that you're talking about. There seems to be a disconnect. Do you have trouble getting young people interested in manufacturing? And if so, why? In some cases. In some cases. I do think a lot of people have bought into the myth that a four-year college degree will be the key to success in the future. And in fact...

there are multiple pathways. So, you know, what I'm encouraged about is the number of young people who are actually starting in high school, maybe taking a youth apprenticeship program, going to the community college, getting exposed to some of the things that, again, I keep describing this as the new tech sector. I think when we get this younger generation interested in the new tech sector, we're going to see a huge acceleration. But, David, it's not just young people.

We used to think that a career was 25 or 30 years and then congratulations, here's the gold watch and you may retire. But I'm past 40 years in my career. That makes two of us. Okay, and I'm hoping to go for 50. And as I look at it, what I'm excited about is bringing more people in from other roles into new fields because, again, because AI makes it possible for us to make that change. That's great.

Now, I know you testified, I think, before a House committee a couple months ago. And I'm interested in your thoughts on how good a job the federal government is doing handling, facilitating, enabling this sort of revolution. Is the government doing enough to position American industry and workers--

to get the skills they need to do this, to compete with China, let's say, at a time when we're cutting research. We were talking to Senator Young earlier today. He said he's been disappointed business has not been more vocal, pushing back on the administration's cuts to the National Science Foundation, university research. This is, as you know, where a lot of the basic research is done in this country that's later commercialized. What's your view of all that?

My view is this. We are in a moment where we know we have to be able to afford our government. So I understand the objectives, right? We've got to right-size our government and get spending under control, no doubt about it. And I think there is a role for business to play. And so what we're working on, and we have with every administration, over 160 years, you can well imagine, Siemens has worked with the federal government throughout our history,

And what we find is that, frankly, national priorities line up with the kind of know-how we have. What I see the US government focused on right now is leadership in AI and tech, bringing manufacturing back to the United States,

the energy that's required to meet one and two, and the workforce that's required to meet one, two, and three. And so we're using our voice to make sure that folks know, hey, here are some tools and techniques that will make it easier for business to play our part. Now let me give you an example of research and development.

globally, Siemens spends 8% of our revenue on research. That's pretty high compared to most companies, but it is our conviction that business steps up, steps to the forefront, does basic research. Not only that, at Siemens, we're big believers that the tools we bring, the digital twin that you mentioned,

We recently invested in Altair, $10 billion investment in a US-based company, Detroit, Michigan, Flint, Michigan, who are very deeply engaged in critical elements of the digital twin that complement ours, so we have the world's most comprehensive digital twin. Likewise, we've just signed with Dotmatics, a company that...

brings those same tools into the early stages of pharmaceutical research and development. Why are we doing this? Because we believe that the private sector in those early phases of research and development can be driving much more of the action. So I think it's a moment where business steps up and we're prepared to do that. - Okay, I'm afraid we're out of time, but I am gonna ask you just very briefly one quick

question about tariffs. You're a global company, you operate across borders, you I'm sure import things that you need for your U.S. operations. How have tariffs affected your operations and how worried are you about all the uncertainty that we still are operating under?

David, Siemens is a global company, but we've been seeing a trend over now decades. And so, for instance, in Siemens USA, we've been localizing. I tell you, 85% of all the supplies we need, our suppliers are 12,000 suppliers in the United States. We've done this all around the world.

That's been an objective. What I've asked my team to pay close attention to is further down in the supply chain, are our suppliers at all exposed? And if so, what we're going to be is very vocal, helping to identify any bottlenecks that prevent us as a nation from achieving our overall objectives of restoring manufacturing and maintaining our leadership in AI and tech.

Okay, well, we'll end it there. This has been a great chat. No questions about trigonometry? I promise those are for later. We'll keep those for later. But Barbara Humpton, thanks very much for joining us today. Thanks for listening. For more conversations like these, follow our Washington Post Live podcast page on Spotify and stay tuned every Friday for our weekly episodes. I'm David J. Lynch for Washington Post Live.