You are using an old browser with security vulnerabilities and can not use the features of this website.
Driverless cars, fully-automated factories, self-learning computer programs: Is technical innovation making humans redundant? Are these developments proof of the fact that, at least with the advancement of artificial intelligence, machines are simply better than we are? No and no, says Gronau. As a business informatics specialist he has researched automated systems and for some time focused on how to improve communication between humans and cyber systems to the benefit of all.
Inside a huge production hall somewhere in the Allgäu region in southern Germany, there is a production line that never stops. It starts with a single motor block. Then, meter by meter, parts are added from both sides and are plugged in, screwed together, or welded by robot arms. At the end, complete tractors are driven from the line. Automated systems like this factory are not new. Robots have been used in industry for decades, and without advanced computer systems urban traffic control would be a problem of serious proportions. Also, for quite some time, weather data have not been modeled with a slide ruler, but with the help of complex mathematical algorithms.
“The problem with many automated systems is that their set of rules needs to be edited and updated continuously if and when conditions change,” says Gronau, who is Professor of Business Informatics with a focus on processes and systems at the University of Potsdam. To identify changes, take them into consideration, and adapt one’s actions accordingly – in other words, to learn – has long been the privilege of humans. However, these days advanced algorithms are being used that are able to independently identify patterns in datasets. Artificial intelligence has prepared the ground for self-learning systems. Very popular, and already in widespread use, are voice recognition systems like the ones used in virtual assistants such as Apple’s “Siri” or Amazon’s “Alexa”. If Alexa has not understood, she asks again and can correct herself immediately. This illustrates that AI needs one thing above all else: a lot of data. “Siri benefits from around a billion users producing several gigabytes of data every day.” In so far, the innovation thrust fits in our time: The collection of big data at many places was instrumental for AI systems to be developed and used in a targeted and meaningful way. The development of individualized medical treatments, the research of old DNA, or the analysis of complex seismic data – they all benefit greatly from the new possibilities. But the dependence on big data also reveals the limitations of AI systems: “A shipyard builds four cruise liners a year, so there is not much for AI to learn from.” And there are many similar cases: As long as correlations, problems, or tasks cannot be converted into big data, AI cannot help process them.
But even now, AI shows its full potential in many ways, and also its superiority: “Machines are always better at taking the right decisions in standard situations and with high precision provided there is a good data basis,” Gronau explains. However, they don’t make humans redundant. Humans are superior where the data basis is poor, individual decisions have to be made, and intuition or creativity are required. Even now, this has significant consequences for the world of labor. Many jobs are done by machines or programs as they are faster, more precise, or more efficient than humans. “Robots as the symbols of digitization do not always inspire a lot of enthusiasm,” Gronau knows. “Many people fear being made redundant by them. But I am sure: Automation creates jobs. And we should remain confident, as cyber systems are our helpers and always will be.” For human labor, digitization has brought about a trend towards de-specialization, Gronau explains. Jobs for which expert knowledge was needed in the past are now being taken over by so-called assistance systems. For instance, compositors are no longer needed by daily papers, as the editors themselves insert their articles into the respective program. Instead, there is an increasing demand for people who are able to plan, control, or develop these systems further and understand a variety of things.
So, how can we make the best use of technical assistants in the future? Gronau has been doing a lot of research on the opportunities offered by autonomous systems. “For a long time, machines and workpieces in factories were not intelligent. A machine processed thousands of pieces, and didn’t know anything about it. The pieces had no idea of what they were, where they came from, and whether they were important,” Gronau explains. But this is changing now. More and more machines, but also workpieces, are equipped with sensors, minicomputers and communication technology. Thus the networked parts of a fully automated factory are able to collect and process data, communicate with one another – and organize, optimize, and also develop a production process almost entirely themselves. Gronau and his team have studied this idea for a number of years and in a variety of projects. It was precisely for this purpose that a plant was set up at the Griebnitzsee university campus. It consists of a conveyor belt, tunnel-like cubes with touch displays and screens, a robot arm, several sensors, scanners, antennas, cables – and some small metal cubes with displays mounted all round it. Initially, the system was called “Leistungsfähigkeitsbeurteilung unabhängiger Produktionsobjekte” (Performance Assessment of Independent Production Objects), or LUPO, for short. Its task was to simulate – as a virtual factory – any number of production cycles. Thus the researchers, together with business partners, were able to design and test productions that did not yet exist, from chocolate to car factories.
In another project, the team assessed the level of autonomy of production facilities of industrial partners. “LUPO enabled us to determine the autonomy of a factory infinitely,” Gronau explains. “The interesting question is: Which level of autonomy is the best?” The above-mentioned tractor factory in the Allgäu was one of the facilities studied. It turned out that its structure – assembling a tractor from motor to wheels – had many fixed settings, and more autonomy could only be meaningful regarding the material feed. In other factories with less structured processes, like the made-to-order production of machines, the simulation indicated that production would become more efficient if coordinated more autonomously by the networked systems.
Over time, the room-sized mini factory has evolved into the “Research and Application Center Industry 4.0”, or RACI 4.0. These days, Gronau researches not only the autonomy of systems, but also the relationship between humans and machines. For him, this relationship is forward-looking for a number of reasons: “Not everything needs to be 100 percent autonomous,” Gronau underlines. “What is much more important is a functioning communication between humans, machines, and workpieces. It would be something like the operating system for a factory. And even though it may still be a long way off, it will happen, and all sides will benefit from it.”
On the one hand, a learning system depends on the feedback of those controlling it. Thus it must be able to express itself and explain how a certain decision was reached. The complex algorithms it uses need to be hidden below an input and output interface that non-programmers can understand, too. At the same time, humans have the task of learning their role in the new cyber physical systems. “Only when humans not only accept the machines, but operate them confidently, can we speak of successful cooperation,” Gronau states. This transformation is what Gronau and his team have been researching in their project “Metamorphosis of the Factory,” or “Metamo-FAB,” funded by the Federal Ministry of Education and Research (BMBF). “The aim is to enable humans and machines of the future to work hand in hand.” In this project, several research institutions – the universities of Potsdam and Stuttgart, Fraunhofer Institute for Production Systems and Design Technology (Fraunhofer IPK) in Berlin – have joined forces with partners from practice.
The Potsdam subproject focuses on humans and their role in Industry 4.0. In a factory with more or less smart technical systems, staff members are no longer button pushers or part loaders. “They become flexibly acting problem solvers,” Gronau explains. Particularly important are their interaction and process competence and self-organizing capacity. The transfer of these competences is a challenge employers are facing even today. This is not an easy task, as in many places factories of the future are not yet available. First, they need to be designed and tested – and this is where RACI 4.0 comes in, as a “learning factory”. “Here employees can train for the workplaces of tomorrow, close-to-process, individually, and with a focus on what exactly they have to learn,” Gronau says. Pilot trainings in cooperation with the Industrial Union of Metalworkers (IG Metall) have made it very clear: Humans can easily acquire the necessary competence to communicate with machines at eye level.
Prof. Dr.-Ing. Norbert Gronau studied mechanical engineering and business economics at Technische Universität Berlin. He has held a chair at the University of Potsdam since April 2004. Currently, his main research interests are business knowledge management and adaptable information systems.
Text: Matthias Zimmermann
Translation: Susanne Voigt
Published online by: Marieke Bäumer
Contact to the online editorial office: onlineredaktionuni-potsdamde