Jim Hagemann Snabe - Doing Good Business

Leadership is about the change you can make happen. It's not about your title. It's not about the resources you have at hand. It's about the impact you can have.

I am fine actually.

Yeah. No, it's actually true. I was quite inspired by math early in my studies. My master's was actually in operational research, trying to solve a very complex mathematical problem called the quadratic assignment problem. And when I left the university, I thought actually at the time that there was an optimal solution to every problem because that's what I had studied, finding the optimum. And I spent a lot of years actually still believing that. But in the meantime, I have understood that the world is more unpredictable. And math, while it plays a role in trying to make complexity understood and even decompose big problems, it does not describe the real world. So when I look at my learnings from math, I guess I'm moving my leadership style away a little bit from being able to predict and mathematically find the optimum solution and more of an iterative learning kind of experience, more actually inspired by biology where you could argue the mutation is the innovation in business, which has inspired me even more the last 10 years.

Yeah. You know, that is back probably what you're referring to, a trip I did in 2008 as part of my attempt to become more of a global leader. I had worked and been a leader at SAP for many years, and I was invited on this journey, partly to understand globalization better but also to see reality. And I spent a week mostly in India, and in a single day I would meet some of the poorest people and some of the richest people on the planet and reflect on topics like what is happiness all about. There was clearly no correlation between wealth and happiness. And I came back with this idea that leadership is about the change you can make happen. It's not about your title. It's not about the resources you have at hand. It's about the impact you can have. And the moment that inspired me the most was a young woman, I think at the time she was 23, and she had found a way to educate the kids in construction sites in Mumbai whose parents live on these construction sites and therefore they cannot go to school. And she had found a way to deliver a school at the construction site with women from the neighborhood and $12 per kid per month. So a good story of how you can have great impact with no resources because you have strong leadership, a big dream, and a big problem to solve, and then make sure it happens. So I believe that business is much more than just shareholder value. It's about creating value in the communities in which we operate. And that's also what inspires me a lot by Siemens. We have this incredible impact in the communities in which we operate. And when we master technology to do that, we can really make a difference in the world, a positive one.

Yeah, I mean, first of all, I think the importance of data cannot be underestimated. In particular, if you add artificial intelligence on top of that. It's funny because when I studied math, at the time it was descriptive models and we did have artificial intelligence classes, but we had two problems at the time. There was not enough data and there was not enough compute power. And then you can't really make AI work. And that has changed since then. Now we actually have a situation where we have enough data. Some would argue too much, and I'll come back to that. And we have compute power that allows algorithms to self-learn from observations. And that creates a whole new opportunity of creating value and of course to create new business models. And I guess that's where your question goes in depth, because what we have seen in the first wave of AI is of course a collection of data, mostly consumer data. I think we have to some extent been luring the data away from people by offering them some free application on the phone without them necessarily knowing what kind of data is being collected and the data is being used in what I call irresponsible ways. Not only do you lose your privacy, which you could argue is one level of problem, but when these algorithms that are designed to persuade you for a certain opinion are used for political opinions or even fake opinions, then it becomes a very dangerous use of data and use of AI. So I have been advocating a lot that what we need to use data and AI for is to enhance human capacity. It is to strengthen our democracies. It is to give people a voice and make meaning out of data so that we can create a better future. And I think we haven't seen that from some companies so far. They have been misusing data and they have been misusing the power of AI to create something that deteriorates human values and potentially kills democracies. And that's why I'm so vocal about it. I am, maybe as a last comment, very proud of the way Siemens is dealing with this because we have set some principles for the way we use data and some principles for the way we use AI. And I think that goes very hand in hand with a purpose-driven company that has progress in society in mind, not just own profits and then never mind whether it's done in a reasonable way like some companies.

Yeah. No, I think that's a very, very important question. Now if we believe like I do that data is kind of the source for the value creation on AI, and I remember maybe an anecdote when I first joined the board of AP Møller-Mærsk, the founding family asked me at the time Mærsk was a conglomerate, they had a lot of business in the oil sector, and the third generation of the family asked me: 'What is our next oil?' And my answer: because I think it's data and smart people. So if you believe that that is true, then of course the more data we have, the smarter we can make our AI and the better decisions we can make with the people we have. And there the sharing of data becomes vital. If we look at the consumer world, I think what we have seen so far is almost a tendency to create what I call data monopolies where there are few companies because of the network effect on platforms. They sit on an incredible amount of data far beyond what we even had imagined. And if that is a monopolized set of data, it becomes an unfair competitive advantage in my mind and it misses out on the societal value that the data could have. Now in the industrial space, we have a very different situation because you can't steal someone's data in the industrial space. You can't tell the wind turbine or the Siemens train or the operations controls in a factory from Siemens that it should share its data in return for something free. It is Siemens who decides who gets access to the data, and of course the customer. And I am convinced that we are at an inflection point where we need to go from what I call ego systems where few people master and monopolize a lot of data and create a lot of value for very few people to eco systems where the data is a shared source of information and we contribute with our data as Siemens because then we get access to other parts of data. And with that we can solve bigger problems for our customers. A good example could be in the energy system space where of course a renewable energy system is very volatile. You don't know when you produce energy and you certainly don't know that it's not when you need it. So you have this unpredictability in the system and AI is the only way to solve that problem. You need to look at patterns, weather reports, and consumption patterns, and then begin to understand how to level that system out. Now if Siemens does that alone, of course we can't manage the spikes in the system of Germany. And so this problem, which is a big problem to solve in order for us to move to renewable energy sources, can only be solved if we all share the data. Now then you can ask yourself, well, is that then a competitive disadvantage? I think not. I think it offers Siemens, if we share our data with others even competitors and get their data of course, it offers us an opportunity to solve bigger problems, hence have more value creation. And then you can't defend yourself from a competitiveness point of view by monopolizing the data. You simply have to be better. And I'm not afraid of that competition. I'm sure we can be.

I think this is the key question around AI. I mean, I've studied the algorithms. I actually took classes on US universities online for the last three years in order to better understand how these algorithms work. And they are super complex. I thought my math when I did my masters was complicated, but the math in AI is very complicated. And the uniqueness is of course that it is programming itself, so to say, it is learning from observations. And with that, it's creating models for probabilities and comes up with recommendations based on probabilities based on the data it has used to learn. In particular, deep learning algorithms have such a pattern. And what that does is in the current version of the deep learning algorithms, there is no real explainability, which means when the algorithm comes back with a response, you have to choose whether you believe it or not. Now, in some responses, in some areas, it doesn't matter. Like Netflix, it will suggest to me which movies I like based on patterns from other viewers. And for that, I don't really care how it came up with that. It may be right, it may be wrong. Who cares? And of course, it gets better over time, but I really don't need to know why it is suggesting some movie. Whereas, if someone is using AI to diagnose me in healthcare, I would really want to know why it believes I have cancer or not. And that's where I think the explainability comes in. Now my key idea for the explainability is that we keep one rule stable, which is there has to be a human being responsible for the final decision, not an AI. And if that is a rule that we apply, then if I'm asked to take responsibility for the AI or the final decision and I get a response from AI and it's a critical decision, of course I would want to know why does the AI suggest this. Hence yes, it's a burden, but I think we need to build it into the algorithms themselves. It's unsolved today, but there are many researchers looking into this problem. And I would not unleash AI in critical areas without an element of explainability and not without a human being finally being responsible. The way I began to think about this was a project actually for the European Union where I was asked to participate in a high-level group of people and we were asked the following question: What should the rules be for weapon systems, autonomous weapon systems? And it's a great question to ask yourself because now it's clearly not the Netflix movie. I mean, you have a machine that has abilities to kill someone. And of course, you conclude there must be a human being responsible. You can't blame it on the robot. And once you have done that, you must have explainability. I think it's key. For instance, we cannot see bias of AI unless we have explainability. So a simple thing like a recruitment AI can be super biased based on historic data and we won't notice because we don't have explainability. We have to solve that problem. So my hope is that we have certified algorithms with explainability elements inside and that companies will use them and not unexplainable AI with the consequences that it might have.

Well, it's interesting. I've seen a couple of those examples where when you copy the system to another location, for instance, it doesn't work. The easiest one to understand is imagine a self-driving vehicle in Germany and in Mumbai where I met the leader I talked about earlier. I can assure you the algorithm in Mumbai has to be very different because a red light doesn't mean people stop the car. They just are a little bit more careful. And so the algorithm needs to take completely different decisions based on different assumptions. Now, does that make it not scalable? I'm not sure. I mean, of course it means more effort, but I actually think it's good effort. I think like human beings, where we are arguing we need lifelong learning, which is basically that we continuously observe in our lives. We get better understanding. We may even change our minds as we come to my age about certain things because we stay open-minded. I would love that to be the case for AI as well. You know, imagine lifelong learning AI. I like that idea.

Yeah, I love that.

Yeah. No, I think that's very, very important and it's kind of, I wouldn't say it's a new aspect, but it's certainly an important aspect of understanding the role of business. You're absolutely right. I mean, I think it was Milton Friedman who said in '72 or so: 'The business of business is business.' And that same year the World Economic Forum was created, or the first Davos dialogue meetings was created by Klaus Schwab and he argued stakeholder, and he said: 'Well, there's more than one bottom line. There's one bottom line for the shareholders, but if you don't have a good bottom line for the employees, how will you make sure that you can even deliver on the shareholders? If there's not a good bottom line for the customer, how can you even imagine delivering any value for your shareholders?' And in recent years, of course, the fourth bottom line is the one for the planet where the industrial world, 200 years of optimization has taught us to maximize outcomes but accept incredible waste and also pollution, which is of course externalities that we should take into the way we look at business. So I'm a big believer in what you mentioned, stakeholder capitalism, which is really about how does business deliver value to all stakeholders, not just one. It was originally invented in '72, as I said, by Klaus Schwab. And many have argued against it because they say: 'Well, if you have to balance, either it's good for the shareholder or it's good for the other stakeholders, then the shareholder always wins.' And that is probably true. I mean, if you're not making profits, then you go out of business and then you can't make anything good for the planet anyway. But I think with technology, and we're coming back now to data and AI and modern technologies, we begin to move these two circles so they overlap so that value for shareholders could also be good value for the other stakeholders. Investing in people so they can use modern technology is good value for employees and I think good value for shareholders. Investing in ways to get customers more value, so they get more value. And you look at our industrialization and verticalization of solutions, that is also good for customers and shareholders. And even on the sustainability, the climate front, I mean, we have more than half of our revenues at Siemens connected with products that help our customers become more sustainable, so pollute less, have less CO2 emissions. And so suddenly it's not like in the old world where you would make business for the shareholders, you would make a profit and then you would have a difficult decision to say how much of my profits will I spend on doing good. So it's not how you spend your money today. It's about how you earn your money. How do I earn money in a way that's good for shareholders, good for customers, good for employees, and good for the planet? And I truly believe that that's possible, in particular when you have tools like AI and data. Because for instance, we can use data from vibrations from a wind turbine and increase the uptime of that system. And because we do that, we do better business with that turbine. It's more productive for its customer and it's better for the environment. And so suddenly we have opportunities where we can optimize entire systems because we have the compute power, we have the algorithms, and we have access to the data in ways we could never do before. And that is what makes stakeholder capitalism possible without too many tradeoffs. And that excites me. And so I always argue that we live in this unique moment in history where it's suddenly good business to do the right thing. And I feel super lucky to be living and leading at this moment of history, but also to do that with Siemens because I can't imagine a company that has more impact on some of the most critical infrastructures in this world than Siemens. Now we apply data and AI to that. Wow, there's so much we can do.

Yeah, I think that's a good way of putting it actually. I think if you believe in stakeholder capitalism, which is about creating value for all stakeholders, then instead of starting with a product where you say 'Okay, I have a product, what can I use it for?' and then you may even misuse it—nuclear power can be misused for other things—I think you start with a purpose, which is the question around which problems am I solving? And if you look at Siemens today, our purpose-driven mentality is one of trying to solve some of the most pressing problems that this planet has. I mean, they are even well articulated. The United Nations has articulated 17 so-called SDGs, sustainable development goals, and Siemens is addressing many of them. So we're trying to make manufacturing that has less waste and is more sustainable. We're trying to make transportation that is sustainable and more convenient. We're trying to make individualized healthcare and even preventive healthcare ideally with our healthcare. We're trying to create urban areas with sustainability and recycling of resources and low consumption of energy. And when you start with that, then you begin to think about which products do I need in order to serve that. And suddenly it's no longer just a piece of hardware, which is not a diminishing of the value of that, but imagine you can combine a piece of hardware with some software that solves that problem in a more holistic way. So then you solve a bigger problem, you have a higher purpose, and your hardware suddenly has a much higher value. And that's why we're seeing Siemens now successful in its focus structure, increasing our growth rates, increasing our margins. It's not because we try to do more things with fewer people. It's because we're adding more value because we're solving bigger problems and because we're combining hardware and software and data and AI in ways nobody has done before. So yes, that is technology with purpose and it's the way to solve the stakeholder capitalism problem.