Interview with Dr. Ulrich Dorndorf
Are you afraid of a chess computer?
INFORM uses artificial intelligent (AI) methods in many of its software products to help companies find the best solution when faced with numerous options and challenged by the task of making highly complex de-cisions quickly. Dr. Ulrich Dorndorf, Chief Technical Officer at INFORM, discusses how INFORM uses artificial intelligence, the role it will play in the future, and the opportunities and risks it presents.
? Artificial intelligence is a buzzword today. But what exactly does it mean?
The term "artificial intelligence" describes the development of software systems that exhibit intelligent behavior. However, the concept of intelligence, in this case, is not used in a broad sense, as it applies to humans. Artificial intelligence is used only in clearly defined fields of application, in which a software system imitates intelligent, human behavior. With the right amount of computing power, these systems use artificial intelligence to achieve results that exceed those of humans. Let’s take a chess computer, for example. Because it can calculate every possible move quickly, and with each move, learn to identify which move makes the most sense, it can play significantly better than a human. Nevertheless, a chess computer can use this specific intelligence only during a chess game and nowhere else. But the fact that artificial intelligence can be used in specialized fields where it is more effective than humans means companies, in particular, can use it to plan and optimize their business processes. And many of these processes would no longer be feasible with any degree of efficiency today without the help of AI.
"Artificial intelligence is used only in clearly defined fields of application. But being more effective than humans in specific areas with the help of artificial intelligence helps companies optimize operational processes."
Dr. Ulrich Dorndorf, Chief Technical Officer at INFORM
? That is why companies, in particular, can benefit from artificial intelligence. What are the strengths of AI in the business world?
In a world where markets are interconnected globally and products are becoming more and more customized with increasingly shorter delivery times, companies must quickly make the best possible decision in every situation. The key to these decisions is data. However, the sheer volume of data and their correlations are so complex that humans alone could never derive the best possible decision. The challenge lies therefore not in the data volume itself but rather in the plethora of alternatives that result from it. Intelligent systems transform data volume into knowledge and furthermore, are able to suggest the best possible decisions. To do this, the systems are equipped with algorithms that use the data to learn iteratively and identify correlations that are barely visible to humans.
______
"The challenge is not the amount of data. The challenge is the abundance of alternative options.”
? The development of these types of decision-making, intelligent systems is at the core of INFORM's business. How are these systems used?
In addition to airport ground handling, our systems are used in transport and transshipment logistics as well as production planning. Furthermore, banks and insurance companies use these systems in fraud detection. Businesses need answers to questions such as:
- How long do I delay the departure of one flight so passengers from a delayed connecting flight can still make their connection?
- Should I consider a last-minute offer over a large delivery of merchandise?
- Can I guarantee the customer an earlier delivery date during contract negotiations?
- Where should a container be temporarily stored until it is loaded for further shipment?
- Is a particular money transaction legal, or is fraud involved?
In order to be able to answer these questions, companies rely not only on the experience and intuition of their employees, but also on software systems with AI.
"From scheduling a fleet of trucks to figuring out the best machine configuration for filling a production hall, companies save huge amounts of money by making calculations based purely on OR."
? Artificial intelligence, as an umbrella term, comprises various methods and technologies. Which ones do INFORM use?
One major technology is machine learning, where useful information can be found in large volumes of data. Algorithms filter out meaningful and useful correlations and can make both predictions and intelligent decisions using this data. This means it is possible to predict flight delays so that ground handling can make adjustments in time.
? You also use methods that aren't at first glance necessarily linked to artificial intelligence.
Correct. However this is only because these methods are older than the term artificial intelligence as we understand it today. At INFORM, we rely on operations research (OR). Developed in the mid-20th century, OR can be considered a branch or complementary discipline of AI today. OR uses methods from mathematics, statistics, and probability calculation to solve business-related problems. Businesses can look at all of the parameters and reliably calculate which processes will produce the optimal yield. From scheduling a fleet of trucks to figuring out the best machine configuration for filling a production hall, companies save huge amounts of money by making calculations based purely on OR. In combination with machine learning, the potential can be even greater.
? Does INFORM also use fuzzy logic for some software systems?
Yes. At the beginning of the 90s, we started to develop the first know-ledge-based systems using fuzzy logic. Fuzzy logic uses rules to re-present human thought in broad terms in software. INFORM uses this technology to help banks and insurance companies detect fraud. In addition to operations research, we have integrated various machine learning methods in our software over the years. These methods differ depending on the area of application and they complement one another.
? Many people are afraid of the "term artificial intelligence." They think that computers will behave intelligently like humans in the future, only better. Just how intelligent is AI?
Artificial intelligence is primarily a niche intelligence. In clearly defined fields, computer systems can yield better results than humans. However, as we discussed, the intelligence of a chess computer is limited to the chess game itself. Unlike humans, the computer is not capable of combinatorial logic and cannot transfer it to other areas. The dystopian fear that artificial intelligence will control humans is something out of science fiction. Or, are you really afraid of a chess computer?
"The dystopian fear that artificial intelligence will control humans is something out of science fiction. Or, are you really afraid of a chess computer?"
? Nevertheless, people fear that AI is starting to replace humans or will make decisions for them.
In the industrial sector today, it is hard to make decisions at the speed and with the quality required without the support of AI systems. Nevertheless, it is still important that humans understand these decisions. Systems that compute solutions in a black box of sorts – one that leaves even the developers in the dark about how the decision was reached – are not well received. We notice in the banking sector that when it comes to fraud detection. Specialists want to understand at all costs why transactions are declined. This knowledge is extremely important for optimizing the system and identifying trends.
? Nevertheless, we can assume that AI will transform the business world. To what degree do you think this will happen?
AI will not make humans superfluous, as is often suggested. However, it is true that AI will play a role in greatly changing certain professions or make them disappear altogether. In return, new professions will be created. These will be accompanied by a new attitude toward technology, as we are already seeing today. An experienced operations manager from the baby boomer generation would surely rate the benefits of chatbots differently than a management trainee born after the year 2000. And the driver of a straddle carrier at the Hamburg port, who was trained more than two decades ago, will most certainly find it harder to carry out tasks assigned to him by an AI system than a digital native. The generation of people born after 2010 knows only a world full of AI systems. And many of the children of this generation will have jobs that do not currently exist.
______
"Humans will never become superfluous!"
? Beyond the corporate environment, what does the future of AI development look like, and what are the greatest challenges we face when interacting with it?
We can assume that AI will find its way into many other areas of both the private and professional worlds. The self-driving car will most likely drive on our roads. But again, this is a case of niche intelligence. For the future, the main question is how can we use this technology responsibly so that we understand the ethical and moral consequences? With everything artificial intelligence can do, humans must always be the focus. A system can offer support, but humans must never give up control. This applies to both the self-driving car as well as all of the other areas of application. The technology itself is morally neutral. That is why it is even more important that we continuously understand who is using which systems for which tasks and why, and on which basis decisions are being made. This social discourse has only just begun.
Mathematical Optimization
It is one of the oldest branches of Artificial Intelligence.
Read here why modern business processes cannot do without it!