act legal Publications
When Robots decide!
Not a new invention of the 21st century, but already since Alan Turing's question “Can machines think?” in the 1950s a term.
The robot as your „boss”
“My boss is a robot”, the employees of the investment company Deep Knowledge Ventures in Hong Kong must have thought when a computer algorithm was appointed as a full member of the board of directors. And in fact, robots are conquering our working world at record speed. They order goods, work hand in hand with human colleagues on assembly lines, distribute products on the Internet and dictate high-frequency trading. Technological progress is enabling employers to use intelligent systems not only to supplement human labor, but also to replace it in certain areas, including supervisory functions. It is therefore high time to deal with the (labor law) legal framework.
It is agreed that the appointment of a computer algorithm to the supervisory board or the board of directors of a German stock corporation is not (yet) possible (sects. 76 para. 3 sent. 1, 100 para. 1 sent. 1 German Stock Corporation Act – AktG). However, the assumption of the supervisory function by a robot is not a future dream. The Japanese electronics group Hitachi, for example, has work instructions executed by intelligent systems in its German branches. Legally, there is nothing to prevent the delegation of the right to issue instructions from a human to a machine if the employer ensures, by means of suitable programming, that the essential circumstances of an individual case are balanced before issuing the instruction and that affected interests are taken into account appropriately (sects. 315 German Civil Code – BGB, 106 German Trade Regulation – GewO). The only apparent limit for the “robot boss” is found in article 22 para. 2 of the GDPR, according to which decisions which have “legal effects” or “significant adverse effects” on the person concerned must not be based on automated processing of personal data. Thus, the robot boss may not (yet) fire. Already reality and after a thorough evaluation in terms of data protection law, it is legally permissible for humans to make personnel decisions prepared by algorithms, especially in the case of fully automated application procedures and selection tools. (“People Analytics”).
So, robots replace workers. But can they also be qualified as employees under labor and company law? Do they count towards the thresholds that determine how many employees are required to install a co-determined supervisory board, how large the works council has to be or whether the German Dismissal Protection Act applies? Are robots to be considered on an equal footing with human employees during social selection? Even if some authors demand this, it must be denied in each case. All threshold values (still) presuppose a natural person and social selection would be taken ad absurdum. Instead of the dismissal of two employees, the “young” robot without alimony obligations is likely to be less worthy of social protection. This would cancel out the constitutionally protected entrepreneurial freedom to optimize work processes by using modern technologies. Robots remain things.
Is the machine „liable”?
Consequently, even humanoid robots are not liable for injuries to employees/customers and/or damage to the company’s reputation that they cause. In order to determine the person responsible, it depends on whether the damage caused is due to incorrect programming or incorrect operation of the robot. The legislator has not yet taken action. It is currently being discussed whether the use of intelligent systems should be subject to the legal reservation of taking out compulsory insurance. The following precautions are particularly recommended: Carrying out risk assessments (sect. 5 German Working Conditions Act – ArbSchG), observance of the safety precautions mentioned in the Industrial Safety Ordinance as well as the ISO standard 10218-2011 and involvement of the works council (sects. 90, 87 para. 1 No. 7 German Works Constitution Act – BetrVG).
Robotics policy required!
But how does a board of directors / supervisory board control self-learning systems that work on the principle of trial and error and automatically change themselves? How does a board of directors act dutifully on the basis of “appropriate information” if it has no insight into the black box that lies between the input and output levels? Company management is expected to balance the potential risks in each individual case, closely monitor developments in the robots they use and implement compliance systems to protect against (legal) violations. It also appears to be legally necessary to lay down the aforementioned legal framework conditions in a written “robot policy”. Company management is well advised to communicate the use of artificial intelligence transparently, especially to investors, employees and business partners. No one is prevented from investing their money in risky projects, working with robots and/or trading with them. But they should know.