Artificial Intelligence – A New Challenge for Technology and Law?

Report about the Karlsruhe Institute of Technologies (KIT) annual symposium “Karlsruher Dialog Technik und Recht“ in December 2018

Dr. J. Schmid, Member of the Editorial Committee

The former judge of the German Federal Court of Justice Prof. Dr. Klaus-J. Melullis heads a research group on patent law at the Center of Applied Law of Karlsruhe Institute of Technology. Prof. Mellulis together with the German Bar Associations academy organizes an annual symposium on current topics at the interface of technology and patent law. The goal of this conference series is to foster a dialogue between the competence fields of law on the one hand and technology on the other. This year’s symposium was directed at the ongoing developments in the field of artificial intelligence (AI) in general and in particular neural networks and machine learning. Among the addressed topics were inventions or creations made by autonomous systems as well as decisions of such systems that can potentially not be tracked back to humans without further consideration. The symposium was attended by about a hundred participants from research and law practice. After an introduction by Prof. Melullis, the symposium featured nine talks by high-profile researchers and practitioners. In the following, an overview of the topics is given.

1. AI in robotics (Prof. Dr. Torsten Kröger)

Prof. Kröger is a lecturer at the faculty of computer science at KIT and works in the field of robotics. He opened his talk by pointing out that it is difficult to find a single definition of AI. For this reason, he prefers the well-defined term of machine learning. He described examples of intelligent (learning) behavior of robotic systems. Further, he discussed technical concepts of machine learning algorithms and showed examples of the abilities of current systems. Still further, he presented possible new application areas of machine learning approaches. One message of Prof. Kröger was that machine learning is mainly a tool and not a magic bullet. This tool, however, has a huge potential to help, support and assist humans at work and in everyday life.

2. New concepts for responsibility in the context of AI and robotics (Prof. Dr. Susanne Beck, LL.M.)

Prof. Beck is a lecturer at the faculty of law at Leibniz-University Hannover and conducts research in the field of criminal law. In her talk she focused on challenges faced by lawmakers with the upcoming of intelligent robotic systems. In particular, she discussed civil and criminal liability of AI systems. Potentially liable entities include the producer of an intelligent system (i.e. the programmer or the vendor) and the user of the system. In addition, she described the concept of the electronic person (e-person) representing the system itself and discussed concepts for fully or partially imposing liability on this electronic person. As an example, Prof. Beck discussed whether it is appropriate to impose liability on the driver of an autonomous vehicle who has relied upon his system for a long time before an accident occurs. She concluded her talk with pointing out that the ongoing developments require an appropriate adaption of standards with respect to civil and criminal liability.

3. AI in patent applications and in the work of the German Patent and Trademark Office (GPTO, Dr. Klaus-Dieter Herrmann)

Dr. Herrmann heads one of GPTOs patent departments. In his talk, he reported on the growing number of patent applications relating to AI. In particular, he pointed out that a total of 37.100 patent applications relating to AI has been filed with the GPTO so far. This number has shown a strong increase recently. He provided a few examples for patent applications relating to applications of AI for controlling gas turbines, for medical diagnosis, for predictive maintenance and for autonomous driving. In the second part of his talk, Dr. Herrmann gave an update on the current efforts of the GPTO to enable a semantic search in patent documents based on neural networks. In the discussion following the talk, Dr. Herrmann provided insights into the GPTOs approach regarding sufficiency of disclosure of patent applications relating to AI. Further, he informed the audience that with respect to determining whether an invention relating to AI is to be considered a technical invention, the GPTOs approach is equivalent to the European Patent Offices approach.

4. Technical and economic consequences of AI (Dr. Jochen Hanisch)

Dr. Hanisch is attorney at law and heads the network development and network management Department of AUDI AG, Ingoldstadt. After a brief introduction, Dr. Hanisch provided an overview of applications of AI in the context of autonomous driving. Then, he discussed civil, criminal and public law aspects of AI. He pointed out the various challenges faced by lawmakers with the upcoming new technologies in particular with respect to the relationship between humans and intelligent machines.

5. Copyright protection for creations of machines (Daniel Schönberger, lic. iur., LL.M.)

Daniel Schönberger is head of legal for Switzerland and Austria at Google Switzerland GmbH, Zürich. He reported on Googles AI principles and responsible practices. He lined out an argumentation why it could be beneficial to grant rights to intelligent machines. Mr. Schönberger gave an overview of the risks of upcoming AI technologies. He emphasized, however, that the fact that innovation ultimately drives prosperity and economic growth should not be neglected. As a consequence, it is required to find and create a legislative environment that allows and incentivizes developments in this field rather than closes down creativity in a misconceived effort to protect it.

6. Patentability of technological developments by AI (Prof. Dr. Andreas Wiebe, LL.M.)

Prof. Wiebe teaches law at Georg-August-University Göttingen and conducts research in the fields of competition and intellectual property law. In his talk, he discussed problems relating to inventions made by intelligent machines. If an intelligent system independently, i.e. without human guidance, can recognize patterns and derive therefrom an independent course of action to develop innovative products, it is possible that such a system invents a new concept. Prof. Wiebe raised and discussed the question whether an intelligent system could be an inventor. One resulting problem then lies in the question of ownership of the invention (user, developer or owner of the intelligent system?). Another problem could arise from intelligent systems generating patent applications or prior art (cf. approaches like Prof. Wiebe recognized the conflict potential with respect to the current patent system resulting from machines becoming more and more intelligent. One possible solution could be to introduce a new separate patent system for inventions made by intelligent systems.

7. Questions in the context of a conclusion of contract by intelligent systems (Dr. Torsten Kraul, LL.M.)

Dr. Kraul is attorney-at-law with Noerr LLP, Berlin and focusses on telecommunication, internet, e-commerce and IT law. In his talk he discussed advantages and disadvantages of different concepts for introducing autonomous systems into the legal framework of the conclusion of contracts. For instance, an autonomous system could either be treated as a representative or as a simple messenger for a natural person. Dr. Kraul discussed implications for contracts resulting from these different concepts. According to Dr. Kraul, there is a need for new regulations covering these upcoming issues.

8. Liability for the consequences of the use of AI (Prof. Dr. Garald Spindler)

Prof. Spindler teaches law at Georg-August-University Göttingen and conducts research in the fields of company, copyright, internet and telecommunication law. In his talk, he discussed challenges for legislation arising from non-deterministic behavior of autonomous systems. In particular, it is a challenge to determine civil and criminal liability relating to decisions involving autonomous intelligent systems. One option to solving these issues is to apply the concept of absolute liability of the person putting the system to use.

9. Digital legal persons: liability for the actions of autonomous software agents (Prof. em. Dr. Gunther Teubner)

Prof. Teubner is emeritus professor of Goethe-University Frankfurt. In his talk, he summarized and discussed risks arising from the upcoming use of autonomous software agents. In particular, these risks can be categorized into (1) an autonomy-risk relating to independent decisions made by software agents, (2) a combination risk relating to a tight cooperation of human and software agents, and (3) a network risk arising from interactions of computers with one another. All three different categories present different challenges for lawmakers and require different approaches. Prof. Teubner provided arguments why the concept of an electronic person might not yet be applicable to current systems.