Assembly robots that build things on their own without having been programmed to do so. Self-optimizing production lines in factories. Trains and wind turbines requesting maintenance based on operational data and artificial intelligence (AI) that can predict behaviour better than the engineers who designed and built the systems can. These developments are a real opportunity if we come up with ideas on how to shape AI and make it a job engine.
It’s beyond question that the world of work will continue to change with the ascendancy of AI. Today, robots still have to be content with the so-called “3 ‘D’ jobs” – tasks that are dumb, dirty and dangerous.
According to recent studies on the future of work, however, this restriction will soon be overcome. By the year 2030, up to 375 million people worldwide will have to learn a new profession. This corresponds to one out of every three employees. And displacement won’t just impact those who perform so-called “simple” tasks, but also lawyers, doctors and engineers.
Forecasts from leading market research companies unanimously confirm that the activities accounting for up to 50% of most tasks can be automated. Machines could not only perform these activities but also complete them better and faster than humans can. The upside? Relieved of the drudgery of such tasks, we’ll have more freedom to assess the results obtained, advise customers and patients, or recognize and foster our employee’s abilities.
Transferring findings from the digital world to the real world
The fear-ridden debate of “man vs. machine” is misguided. If you take a closer look, it’s already clear that developments are moving in a different direction. In fact, when we speak of AI today, we’re always talking about the development of artificial intelligence by humans.
We can think of AI as a “black box”: we put knowledge in the box; then a little bit more knowledge; and a little bit more. But in the end, we can’t currently get anything out of the box except for what we’ve put into it. So, “thinking outside the box” will take on a whole new meaning in the age of AI. For us, there’ll be two key developments:
⦁ First, there are still limits constraining AI’s progress. But in the future, machines will increasingly learn independently. They’ll become capable of “thinking outside of the box”, so to speak.
⦁ Second, it is understandable that the increasing involvement of AI in our lives may arouse fears and anxieties and we must take these fears seriously. But we always have to keep in mind that just as humans have guided and driven the development of AI to this day, they’ll continue to lead in the future.
Just one example: although Deep Blue won against Garry Kasparov in 1997, it’s still taken until now to develop artificial intelligence that can fully exploit its potential in strategy games. Today, neither humans nor “conventional” AI can beat the AlphaZero machine – neither in chess, nor in games of even greater complexity, such as Go or Shogi. However, this development never would have been possible without human intelligence to design this system’s architecture. This illustration shows that it’s not about man vs. machine – it’s about man and machine.
Today, it seems groundbreaking that virtual assistants can make an appointment with the hairdresser for us, or that AI can automatically place orders for us online. But is that really such a valuable innovation? And who really benefits from it? In the end, it’s primarily the e-commerce or marketing platforms themselves that benefit. But, there’s a form of AI that we can all benefit from: AI that’s integrated into industrial processes to create value on an industrial scale.
This change has already begun. To make it happen, industrial enterprises are striving to attract and lure away each other’s talent. The crucial people are the ones whose expertise bridges areas that were once separate disciplines: data scientists who have additional knowledge of physics or engineering, for example. Such individuals are in great demand because they’re the only ones able to translate AI-generated data about a train, for example, into real-world benefits.
Once data has been translated, railway operators can receive direct instructions about which parts of which railcar must be exchanged by what time. Predictive maintenance, risk analysis, knowledge of spare-part availability and of the legal conditions of the country in which the train travels have already been incorporated into this AI-enabled statement. Only human beings have this ability to translate the underlying insights from the digital world into the real world in a way that puts AI capabilities to good use.
That’s also why we rely on craftsmanship and skilled manual labour and will continue to do so, when these tasks are performed at the highest level. In the manufacturing and maintenance of locomotives at the Siemens factory in Allach near Munich, Germany, mechanics and welders do high-precision work to the tenth of a millimetre. Train availability can only be guaranteed if these jobs are done by professionals who gain their insights from AI.
Taking Industrie 4.0 to the next level
Currently, three developments are happening in parallel. New jobs are being created, outdated jobs are fading, and many of the remaining jobs are changing.
In order for the balance sheet of AI applications to be positive, companies – including large corporations, small and medium-sized businesses, and even tradespeople – must be able to deploy AI in a broad and profitable manner.
I’m talking about industrial AI – the combination of AI with domain know-how. Siemens aims to establish “digital companions” that act as a kind of human-intelligence enhancer.
This type of AI, one that supports us as humans, needs to be widely accessible. However, making this technology available not only requires investments in research and development, but also a much sharper focus on education and skills development.
At Siemens, we spend more than half a billion euros on training each year, and digital skills are part of all our training programmes. However, in the end, skills development has to start much earlier: the acquisition of skills must start at a basic level in pre-schools, continue in elementary and high school, and eventually deepen and specialize at universities.
Successfully shaping the Fourth Industrial Revolution and remaining internationally competitive in the digital age will require great efforts on the part of leaders in industry, politics, science and labour organizations.
Germany’s successful Industrie 4.0 initiative offers a good example of the kind of fruits such joint efforts can bear. Comparable initiatives and networks in other countries have been developed to promote local value creation and digital transformation in industry. Examples include Made in China 2025, Manufacturing USA, Make in India, and Egypt Vision 2030.
Industrial AI can give the Fourth Industrial Revolution a huge boost and take Industrie 4.0 and similar initiatives to the next level.
Outpacing coincidence – boosting productivity
According to recent studies, AI has the potential to increase global gross domestic product (GDP) by an average of 1.2% per year over the next 12 years. Thus, the gain enabled by AI is poised to exceed the 0.6% growth effect that the revolutions that were enabled by steam engines and by the diffusion of information and communication technology achieved in their times.
Looking at the development of global GDP over the last 200 years, we see a trend that has, in recent years and decades, taken an increasingly exponential course. One key reason for this acceleration is technological progress.
Technology has a direct impact on social development, economics, productivity and growth. And especially in the recent past, we’ve seen a tremendous accumulation of innovations and technologies. This is quite surprising because, by definition, innovation is “inefficient.” By that I mean that many innovations and pioneering technological developments were created by accident – just think of the invention of the microwave oven, the development of Teflon or the discovery of X-rays.
Apart from these examples, however, innovation is fundamentally based on the trial-and-error principle: adopting a hypothesis – setting up a test – testing – identifying an error, and so on. The inefficiency of this method means that you often have to launch a series of attempts before you reach your goal. This weak spot is exactly where AI can have enormous impact.
The keys to success in the digital age are speed and scale. And if there’s one area in which AI is already far ahead of us humans, it’s the tremendous speed at which models process data and then detect and exclude errors. In short, AI has the potential to help us avoid mistakes and overcome coincidence. Keeping this in mind, it seems all the more understandable that the McKinsey study “Notes from the frontier: Modeling the impact of AI on the world economy” concludes that AI will add $13 trillion to global added value by 2030.
AI with domain know-how
Siemens currently plays a pioneering role in industrial AI because we recognized the signs of the times early on. Initial successes of our AI experts date back to 1995. Combining AI with domain know-how has changed our offerings.
Unlike the methods used by railway operators, for example, our service offerings don’t simply track down and repair broken parts. We guarantee levels of availability that enable trains to compete with planes – and win.
One example here would be the high-speed rail line between Madrid and Barcelona. The train operated by the Spanish state railway Renfe takes two and a half hours, while the pure flight time is one hour and twenty minutes. If a delay of fifteen minutes or more arises, train passengers get their fare completely refunded.
To ensure high reliability, Renfe has formed a joint venture with Siemens that services the trains with the aid of advanced, AI-powered data analysis. So far, only one out of every 2,300 journeys has incurred a significant delay that was caused by technical problems. The result? When the train line was put into operation a good 10 years ago, Renfe says only 20% of travellers chose to travel by rail; today, it’s more than 60%.
Siemens now employs around 800 experts for data analysis and AI. In recent years, they’ve made many AI-based successes possible in industrial environments, for example:
⦁ With industrial services based on continuously running algorithms integrated into production processes. Thanks to the ongoing collection and analysis of process data, we’re able to continuously retrain machine models and increase the accuracy of predictive analysis. These advances make it possible to reduce costly quality testing – such as X-ray inspection – by more than 30%;
⦁ With algorithms that automatically analyse the operating data, environmental conditions and component properties of gas turbines. This approach extends maintenance intervals by 30% and reduces costs by 16%;
⦁ With healthcare AI that enables Siemens to support doctors in evaluating thousands of X-ray images, and thus in ensuring more reliable diagnoses and better treatment for patients; and
⦁ With extremely complex AI-based quality control capabilities for steel mills. This self-learning system is now a classic offering; since 1995, it’s been installed in 30 steel mills around the world.
What’s more, AI makes it possible to build new business models. Take the Renfe example: Instead of offering spare parts to the customer, Siemens sells train uptime – or the amount of time the train is operational.
There’s a comparable model for companies that make machine tools. AI enables them to greatly enhance their ability to analyze and predict machine wear and tear. This information helps establish a use-based business model for machines.
Leading the digital transformation to success
In the current discussion around AI, one aspect has not been receiving the attention I think it deserves. While AI will be decisive for the further development of GDP, economies, too, must transform in order to adapt to the digital transformation. These changes require a working population whose activities are not primarily labour-intensive, but skill-intensive. Value is created through skills and productivity.
Societies that want to assert themselves in this world must get their economies into shape. Forecasts from leading market researchers are in line with the assessment that AI technologies – used correctly and consistently – have what it takes to boost the gross domestic product of economies.
We currently stand at a crucial point in AI’s evolution: we’re on the verge of breaking into exponential growth. And there’s still plenty of untapped potential.
With Industrie 4.0, we’ve successfully started the digital transformation. With industrial AI, we can now take it to a whole new level. We can outpace error and coincidence. We can drive innovation. We can increase efficiency and productivity. We can shape technological and social progress.