Prof. Dr. oec. HSG Holger K. von Jouanne-Diedrich has been Professor of Business Informatics and Customer Relationship Management at the Aschaffenburg University of Applied Sciences in Bavaria since 2013. He holds a doctorate from the Institute of Information Systems at the University of St. Gallen, Switzerland, and studied business administration at the University of Hamburg. He has also held various management positions at Lufthansa, Deutsche Bahn, Siemens and Atos. His main focus in research, teaching and consulting is in the areas of data science, machine learning, artificial intelligence and quantitative finance, covering both the mathematical-technical and management aspects. He is the author of various articles and the OneR-Machine Learning R-extension package.
I had the pleasure of meeting with him to ask about artificial intelligence and the best ways for companies to innovate. Here is our discussion.
How do you assess the current state of artificial intelligence technologies?
Holger von Jouanne-Diedrich: I see it pragmatically. All analytical technologies, whatever we call them – data science, data mining, machine learning or artificial intelligence – are ultimately tools, i.e., means to an end. Technology alone does not define a new business model. It is simply not the case that data is “thrown into” a process and miraculously new ideas and value-added chains emerge.
However, it is quite impressive how much progress has been made in those technologies: AlphaZero can learn chess at a superhuman level in a few hours. The algorithms have been refined in recent years. Thanks to higher computing power, much larger neural networks can also be trained, which in fact often produce very impressive results.
Nevertheless, this is not yet a generalised intelligence, but rather silos of use cases that will have to be adapted over the next few years.
The problem is that everything basically still has to be custom made. With a lot of experience and high expertise, analytical procedures can be built, developed and applied, but only for one area at a time. We still need superman-like experts in the field of data science: specialists who are experts in both data management and analytics who can discuss business models and communicate the whole thing in an understandable way. Progress in the future will focus on simplifying, automating and industrialising the development of new systems.
But how can new business models be found? What is the best way to innovate?
von Jouanne-Diedrich: There is no general recipe that fits everybody. If we had that, that's what everybody would do. I'm observing that it's not uniform. Every company has to find its own way.
I think we can learn from the experiences of failed companies. Why haven't market leaders like Nokia and Kodak made the next leap in innovation themselves? Nokia laughed at the iPhone, and Kodak kept its blueprints for the digital camera under lock and key.
Many companies are only able to perfect and optimise their already-known processes. But they fail to bring new innovations into the market. But by now everybody should know: “Cannibalise yourself before others do it.”
What role can innovation labs play in this context?
von Jouanne-Diedrich: My impression is that many innovation labs are misplaced. Often the only reason for their existence is the fear of missing out on something. Then they will remain a fig leaf, according to the motto: “We are also present in the hip startup scene in Berlin.”
In my opinion, it is absolutely necessary to have the backing of the company's top management. This must not only be formal, but visible to everyone in the company. However, the exact procedure is an art. A simple “develop what you want” is very uncertain; however, a precisely formulated order contradicts the idea of innovation. What is needed is a method that lets the mind wander and enables ideas to be channelled quickly. Here's the famous Google example, according to which Google employees are allowed to invest a certain amount of time in their “own” projects, which of course must have something to do with Google products in the end.
And of course, luck is part of finding the right use case. A single mega-success will then be able to overcompensate for many attempts. However, these attempts must also be made: “Fail fast” as a strategy must therefore be permitted by the corporate culture. Making mistakes must not be a flaw for the further career, but a condition.
Keyword “corporate culture.” How can labs help change that?
von Jouanne-Diedrich: I think it is questionable whether labs will now really be able to induce a cultural change. The fear of failure is often firmly anchored in the company's DNA. In case of doubt, you therefore avoid risks and forego opportunities. If failures are only allowed to occur at one specific point in the company, namely in a lab, this tends to reinforce the impression that innovation is more likely to be outsourced and that everything in the company's core area should continue to do business as usual.
In my opinion, the popular visits to Silicon Valley don't work either. Of course, all senior managers come back with shining eyes – as after every motivational seminar – but putting new ideas into production requires more profound changes. It is a question of training and career paths, as well as the visible release of hierarchy and an intensified effort for networks of cooperation.
Keil: How does artificial intelligence affect companies and ultimately our society?
von Jouanne-Diedrich: We see a lot of uncertainty here. Recently, there have been frequent reports of imminent job losses, and this has been illustrated with striking examples. Of course, e.g., chatbots answer questions around the clock without becoming tired or demanding days off – and in many cases, they respond well enough to customer enquiries. This causes a great deal of uncertainty. Suddenly one is no longer certain of one's own competence, especially since the factor of speed is always factored in: What may not work well today may become reality tomorrow.
We as a society must deal with this uncertainty. I also don’t have all the answers when it comes to managing the challenges that result when whole occupational groups are rationalised away. It is important to note that this now also affects much more complex and better paid activities, in contrast to the latest waves of automation, such as in the case of banks using ATMs and online banking.
Lawyers and patent attorneys, for example, do a lot of “text mining,” as one would say in machine learning, on the basis of legal texts and knowledge of similar cases and judgments. This can often be replaced in large parts by artificial intelligence. The assessment of MRIs can be automated by image recognition algorithms and thus penetrates the core competencies of radiologists. The best investment strategies are less and less developed by experienced financial market experts, but are optimised by robo-advisors in milliseconds. Services such as deepl.com often achieve the translation quality of professional translators – including nuances from the context. (In fact, large parts of this interview were translated into English by that very service!) We don't even need to talk about the travel and transport sector, but even apparently creative activities such as that of chefs are the focus of this revolution that is coming upon us. These are just a few examples; new ones are added daily.
Even if these professions do not immediately become completely obsolete, the combination of human experts and AI often triggers such a productivity boost that considerably less human labour is needed. This will put our society to the acid test, personally, economically and, of course, also politically. This is not a question for me and can already be observed today.
