Datafication – Or Why we Should Learn to Love our Robots

If there is one word that describes today’s world then perhaps it is the word “digital.” Everything is somehow digital or influenced by the digital. And if there is one concept that explains what digital means, it is “transformation,” since the digital changes everything. But what is digital transformation? Here again, I offer one word that could summarize (almost) everything that can be understood in terms of digital transformation: “datafication.”

Datafication has to do with data. This is obvious. However important data may be, what is more important is what is done with the data, the ways in which the data is handled. This can be called analytics. This sounds a lot like big data analytics or business analytics, and datafication is indeed related to these practices. But the idea behind datafication goes further and encompasses more than big data or business analytics. Interestingly, the idea behind datafication is not new and not even originally digital. When NASA built its space capsules, they did not just build one, but always two. One they sent into space and the other stayed on earth. The idea was that when a problem occurred in space, engineers on earth could attempt to replicate the problem with the model and after finding the best solution, tell the astronauts what they should do. This makes sense, since in space there was not an extensive team of experts nor opportunities available to try this or that solution to see what happens.

Today there is no longer any need to build a physical model since one can create a digital model or a so-called “digital twin,” or “digital double.” This can be any machine, not only a space capsule, for example an automobile, a heating system, a refrigerator, or whatever. One can do much more with the digital model than with a physical model. As with the physical model it is possible to represent all states of the machine and also represent problems that arise and look for solutions. This is the first step and can be considered a particular kind of analytics, namely, descriptive analytics. But because of sensorics and wireless networks it possible non only to gather real-time data about the states of the system, but also about everything that is going on in the environment of the system. Via simulation one can then change any internal or external variable and see in advance what will happen. Contrary to the physical model, digital modeling of the system and its environment mean that we no longer need to wait until a problem actually occurs in order to start looking for a solution. We know in advance what problems will occur under what conditions. This is a second kind of analytics, namely, predictive analytics. Predictive analytics allow us to know, for example, when a machine needs to be serviced or when a part needs to be repaired or replaced before a machine malfunctions.

In addition to descriptive and predictive analytics there is a third form of analytics that can be called preventive analytics. When we know what problems could occur and what variables cause the problems, we can take preventive measures so that the problems don’t occur. All this cannot be done when working only with a physical model. This is only possible on the basis of datafication. Descriptive analytics tell us what is happening, predictive analytics tell us what will happen, and preventive analytics tell us what to do, when we want to change the situation so that problems don’t occur.

Datafication, of course, is not limited to representing machines. Not only machines, but also the entire factory that produces the machines can become a digital twin and can be subjected to a descriptive, predictive and preventive analytics. This is what business analytics to a certain extent has long been doing. Datafication, however, takes evidence-based management to a higher level, since it allows not only problem-solving with regard to certain indicators, but scalable, real-time, ongoing monitoring and optimization of all processes and networks that in any way condition an organization. Datafication can be applied not only to factories or other businesses, but to entire cities, which then become “smart cities.” Datafication encompasses even individual human beings.

It is possible to produce a digital twin of anybody. The genome, the microbiome, epigenetic factors, all vital functions via wearable tracking devices, medical history, environmental influences, etc., everything in any way related to a person’s health or well-being can be datafied, aggregated, and subjected to a descriptive, predictive, and preventive analytics. This can be called “personalized medicine.” Personalized medicine eliminates the difference between being healthy and being sick, since one is always in some ways ill or in the process of becoming ill. Medical care no longer comes into play only after one has developed symptoms. Instead, medical care becomes an ongoing activity that is involved in all aspects of life, including interventions with regard to life-style, work, hobbies, eating habits and preferences, sport and fitness, and even how we arrange our houses and living spaces and what environmental influences we are subject to. This encompassing monitoring and evidence-based decision-making is of course not limited to health care. Datafication will transform education (see for example learning analytics), work, and many other activities. Datafication means that we will no longer make decisions about how we live, what we eat, what profession we learn, where we work, what partner we choose, etc. on the basis of emotion, intuition, habit, or preference, but on the basis of evidence.

This brings us to a fourth form of analytics, otherwise known as artificial intelligence and robotics. This can be called prescriptive analytics. When my Tesla drives me home in auto-pilot mode, then it makes the decisions. These decisions are binding for me and in this sense prescriptive. They prescribe whether I turn left or right, whether I go fast or slow, etc. What makes AI into AI is autonomous learning and decision-making. AI is no longer a mere tool in the hands of humans. It no longer merely provides information or makes recommendations; it acts with its own goals and its own ability to make decisions and is therefore much more a social partner than a tool. This can be seen in a resolution of the European Parliament (2017) that proposes to grant AI’s a recognized and legally anchored “electronic personality.” As AI becomes poised to take over many human activities and put people out of work, many attempt to insure a place for humans in the economy by emphasizing emotion and leadership skills, since it is supposed that these are things that computers can’t do. Leaving rational intelligence to the machines, humans retreat to emotional intelligence and the claim that there should always be a human in the loop to make the final decisions.

Robotics can also be subsumed under prescriptive analytics. The hardware and software change, but the basic idea of datafication and analytics remain the same. Robots are mobile AIs. A robot is an intelligent, mobile, autonomous system, IMAS for short. Hollywood has created an icon for robots, the Terminator. Although the Terminator was at first negative, in subsequent episodes of the story, the Terminator became a hero instead of a villain. Today’s robots no longer look like the Terminator, rather they take on many different shapes and purposes depending on what kind of work they do in industrial production or logistics, autonomous cars, military, etc. And when robots do look like humans, they are usually female, such as Sophia from Hanson Robotics, or Erica from Hiroshi Ishiguro, or Jia Jia from the University of Science and Technology of China. There seems to be a tendency to feminize robots and it is also apparent that humanoid robots are seen very differently in Asia and in the West. Sophia is obviously a machine, whereas Erica and Jia Jia are intended to resemble humans as much as possible. Studies have shown that people trust and confide in robots as much, if not more so, than in humans. There is also little evidence to support preferences for the advantages of human decision-making over that of AI’s. Perhaps the meaning of digital transformation and datafication is not that humans must struggle to find their place in a world where decisions are increasingly evidence based by emphasizing differences and uniquely human characteristics, but by means of emphasizing similarities and cooperation in trusted networks of humans and non-humans.