by F. J. Radermacher
   
Director of the Research Institute for Applied Knowledge Processing (FAW/n) and Professor of Informatics, Ulm University, President of Senat der Wirtschaft e.V., Bonn, Vice President of the Ecosocial Forum Europe, Vienna, and Member of the Club of Rome

Correspondence address: FAW/n, Lise-Meitner-Str. 9, D-89081 Ulm, Germany, tel. +49 (0)731-50 39 100, fax +49 (0)731-50 39 111, e-mail: radermacher@faw-neu-ulm.de, www.faw-neu-ulm.de

Digital machines and the world of algorithms

In recent years, we have seen more discussion about the possibility of artificial intelligence, the power of algorithms and the question of the future of this technology. The recently deceased former editor of the Frankfurter Allgemeine Zeitung, Frank Schirrmacher, expressed concern in his book, "EGO - The Game of Life", about the power of algorithms, discussing the great danger to civilisation.

It is indisputable that the algorithmic processing of information is becoming more and more important. Satellite communication, navigation systems in cars, micro trading on stock markets - these are all governed by algorithms. Algorithms also play a role in describing technical components or houses with CAD models, deriving processing models from these descriptions algorithmically, which we then send to a milling machine or printer, which then manufactures parts, or 'prints' them, as we sometimes say nowadays.

It is no exaggeration to say that the technical potential of a civilisation is determined by the performance of the algorithms it has at its disposal. Practically all machines are embodiments of various (mathematical) algorithms.

In mathematics and theoretical computer science, there is a good understanding of what can be handled using algorithms. In the broadest sense, this is the world of computability. The incarnation of the calculation engine is a computer equipped with the necessary peripheral components (e.g. keyboard, scanner, printer etc.). The mathematical structure that abstracts the computer is the Turing machine. This is the abstract, and conceptually, very simple, and yet maximally powerful machine in terms of performance, operating on an endless write and read tape.

If what we usually perceive as intelligence has a lot to do with having access to relevant algorithms and the ability to execute them, the increasing penetration of the world with abstract, coded algorithms and with machines that can execute these algorithms, constitutes a dominant civilisatoric process.

Today, we are faced with the question of how we should assess this development and how it might look in the future. Obviously, the rapid innovation in IT and robotics is having a huge impact on our lives. "More and more intelligent machines, and in the future, more "human" robots, could make more and more useful services possible, but in the long run they could endanger our jobs, spy on our private lives, pursue us with targeted advertising and almost completely occupy the capacity of our consciousness with social networks.

We humans are becoming more and more transparent, because this is what our employers want, it is what our suppliers and customers want, it is what the state wants and it is what technology can deliver. Meanwhile, we are confronted with the many demands for presence in software-supported social networks. This creates great social pressure to be present in the networks. Our many internet presences leave footprints. Information, particularly metadata about our connections, are systematically stored. This allows our lives to be reconstructed piece by piece at extremely low cost. With Big Data and increasingly powerful algorithms, systems know more and more about us.

What will happen in the future. Today, when we are working on such systems, and doing so in a technically intelligent way, there is still an off button. Will that also be the case in the future? Or will machines control our lives? This would be critical. Because we humans have an emotional hormone-driven assessment system. This includes what some philosophers describe as qualia. This is linked to notions of a just, successful world, emotional reactions to the question of what is right and wrong and the intentionality of striving for a 'good' world. The machines we are talking about here have none of this; they live in a world of symbols only, of words, images and models. The main underpinning of words is, in turn, words. For us, this is different. For us, symbol grounding via the body (including qualia) is the crucial extra variable. So in the qualia question, there is the principal difference between humans and machines, and this will continue to be the case. For interactions with machines, the important thing is what the human builds into the humanoid, what the human makes it capable of. If we create it with intentions, if we build more and more abilities into it via software (including the ability to learn), this becomes potentially more and more dangerous - with or without qualia. This is where we should be clever, restrain ourselves, and experiment with small steps - and always with an off button.

This article is a short version of a translation of a contribution to the book "Die Menschine" FESTO AG, excerpt from FAW report "Algorithms, machine intelligence, BIG DATA: Some basic considerations". An abridged version of the text appeared in the special issue "Big Data versus large data collections. Opportunities and risks for health research" by the Bundesgesundheitsblatt, 2015