What must the German economy do to reposition itself for the AI era? Prof. Dr.-Ing. Rainer Stark sees the key to securing prosperity in industry’s untapped data reserves, trust-based collaboration and new alliances between engineers and computer scientists.

He is Head of the Industrial Information Technology Department at the Technical University of Berlin and Deputy Spokesperson for Science on the “Industry 4.0 Research Advisory Board” of the Academy of Technical Sciences (acatech).

Professor Stark, can you think of any successful AI that is ‘Made in Germany’?

There are certainly some AI applications in industry, for example in development, production and maintenance, as well as AI-based analytics, such as in medicine. In these areas, companies use their own datasets to train AI and make decisions more quickly and reliably. However, these are in-house solutions that are not freely available on the market. In terms of a generally available AI, nothing springs to mind off the top of my head.

Why is that?

There are many reasons for this, starting with access to data. In principle, we have a very large pool of data in industry. There are many large datasets available, but they are often buried in ‘data cellars’ – that is, in separate databases and servers. They are difficult to tap into, as the individual specialist departments rarely discuss sharing their data treasures with one another. IT, for its part, knows nothing about the content of the data collected in production or development. The information on where and in what form data is available would first need to be organised in order to achieve efficiencies within companies or to train AI models. This is where the Chief Digital Officers in companies would need to step in, but they apparently aren’t getting sufficiently involved or lack the necessary detailed knowledge. Another challenge lies in the question: who is willing to share their knowledge with whom so that they can benefit from it together?

Why is this a challenge?

Take, for example, the training of larger AI models that are tailor-made for applications such as concept development, series development, robot simulation or the operation of machines in the factory. At the moment, our approach is rather fragmented. We have a wealth of know-how data within the SME sector, and many are initially trying to train AI models on their own. Logically, however, such models should be developed beyond the scope of individual companies. This is because these systems require as much input as possible from a variety of fields. We need greater willingness to collaborate with supposed competitors from time to time. Many firms still operate in a very insular manner, but there are opportunities here. It doesn’t have to be the latest, most important project for which data is shared.

What would be your message to German industry?

Be more daring and act collaboratively and in a spirit of trust. Both within industry and with the research sector, so that we don’t just try things out on an ad hoc basis, but can quickly put new ideas into practice with strong momentum. In fact, other nations envy us, saying: “You have outstanding research and development and a well-positioned industry.” But too little of this is actually put to operational use or integrated with one another. We had a better handle on this with traditional technologies. In the digital industrial sector, much more needs to be done now. So far, we’re still thinking too much in terms of individual projects. There are collaborative projects that run for several years; you learn a fair bit from them, but then things fizzle out because too many companies think: ‘I’ve now learnt the methodology and can implement it using my existing infrastructure.’ But we’re now entering a new era in which the infrastructure doesn’t even exist yet – for example, to train large AI models. This requires energy, data centres and the data itself. It’s a new mix for which we haven’t been well prepared so far. The game is only just beginning, and we’ll have to work twice as hard.

Why is Germany struggling so much with innovation in this area?

One of many reasons is certainly demographics: we have business leaders with a strong technical background who are often cautious about tackling new things. They try to stick to their existing model for success, and the baby boomer generation has only recognised the need for change to a limited extent. After all, things have worked out so far. In many companies, careers have continued to be built along more traditional lines, rather than on the basis of new ideas and technological expertise. That needs to change.

How could this dilemma be resolved?

One successful approach could be to pair an experienced manager with a wealth of expertise with a promising young talent – someone with plenty of fresh ideas – so that the two of them can develop new ideas together. I believe that would be a wise move, because it would also give the next generation the assurance that: ‘You will be leading this one day; you can benefit from it, but you must also rise to the challenge now of helping to shape this future, so that something new can emerge.’ There is also another important point: we have allowed a divide to exist between engineers and computer scientists for far too long. That was unwise. Engineers have always had a bit of an air of superiority – after all, for decades they were the pride of German industry. And now, in recent years, that prestige has crumbled, and a great many engineers are frustrated because their achievements are no longer really recognised. And computer scientists have been preoccupied with their own topics for too long and now admit: “We need to work much more closely with engineers, because it is precisely this combination that makes it so interesting.” That is now our most important task: bringing these disciplines together, alongside unlocking hidden data.

Why do you see the convergence of engineering expertise and computer science as a major opportunity?

There are actually three points. On the one hand, with the help of heavily data-driven models and simulations, we can take traditional engineering development work to a new level. On the other hand, IT can make the operation of technical systems smarter, better and more sustainable. One of our research projects, INSPIRE, for example, focuses on hybrid decision-making – in other words, combining human expert knowledge with the capabilities of AI and data analysis. The hope is that this will enable us not only to make existing systems more efficient, but also to bring entirely new products to market more quickly. This is also the third point: with the help of artificial intelligence and digitalisation, we’re unlocking a new dimension of value creation and can launch completely new offerings on the market.

What sort of offerings do you have in mind?

There are many possibilities. Let’s say I’m a start-up founder and want to focus on my digital expertise when designing products, but I need reliable implementation in terms of production technology, which I can then obtain as an online service from another company. Or large-scale equipment such as a gas turbine, where customers can add on further functions that do not necessarily have to come from the manufacturer itself. With such offerings, we could also give a boost to the EU’s Green Deal. We need new solutions and must focus on smarter systems if we want to hold our own on the global market with ‘Made in Germany’. Our entire operation of technical systems is now transitioning to a new, networked and autonomous form. We must master these processes – ideally so skilfully that we can do better business with them than others.

The interview was conducted by Karsten Lemm. Contact

Contact: Prof. Dr.-Ing. Rainer Stark; Head of the Industrial Information Technology Group; Institute for Machine Tools and Factory Operations (IWF); Faculty V – Transport and Mechanical Systems; Technical University of Berlin; Email: rainer.stark(at)tu-berlin.de