Artificial intelligence (AI) has the potential to change the working world more profoundly and in more ways than any other technology to date. Its use in the working world is no longer a distant possibility, but a reality. AI applications promise objectivity, higher job satisfaction and employee retention, and above all: efficiency.
They cover the entire cycle of working life, from recruiting and personnel management to long-term personnel development. Organisation and management tasks are already being taken over by automated decision-making systems in many places. We have already written about human resource analytics, which is aimed at hiring employees and evaluating their performance, here. Another clear use case is "algorithmic management" – increasingly, digital systems are now giving employees work instructions which would previously have been issued by the employer within the framework of his or her right to issue instructions as per section 106 sentence 1 German Trade, Commerce and Industry Regulation Act (GewO).
What is algorithmic management?
Algorithmic management includes staff scheduling and the coordination and monitoring of staff activities. This form of AI is in the process of changing the way company structures work.
For example, employees in the logistics industry receive work instructions generated by an algorithm via a digital receiving device. In warehouses, so-called order-picking trolleys guide employees from one storage location to another and give instructions as to what product and what quantity of items should be taken from a particular station. In passenger or goods transport, AI decides on the route the employee driving the vehicle should take and indicates petrol stations, break times and break locations. In retail, AI systems analyse data to predict customer demand and make decisions about the most efficient deployment of staff.
AI as an instructor
The employer's right to issue instructions pursuant to section 106 (1) German Trade, Commerce and Industry Regulation Act (GewO) establishes the personal dependence of employees and is a characteristic feature of employment relationships. However, employers are not completely free to exercise the right to issue instructions: They must always decide at their "reasonable discretion", meaning that they take into account the circumstances of the individual case and the legitimate interests of the employee.
In practice, the employer's right to issue instructions or to exercise it is often transferred to lower-level managers. The legal basis for this is found in the right to issue instructions itself. The work then includes, among other things, exercising the right to issue instructions to other employees on behalf of employers.
But is it also possible to transfer the right to issue instructions to AI systems? German employment law does not establish a principle requiring that instructions and decisions are to be made by a human being. Moreover, since the instruction given by an AI system in any case observes conditions predefined by the employer, employers can delegate their right to issue instructions to an AI system. The instructions given by the AI system are then attributable to the employer.
However, if an instruction is deemed "unreasonable", the employee does not have to follow it. In cases of doubt, employers must explain and prove whether a decision was "reasonable". In this context, the question arises in particular as to whether AI instructions could be unreasonable per se because no comprehensive balancing of interests takes place. AI instructions are usually based on the data accessible to the AI or collected by the AI itself. For the most part, an AI system does not know the circumstances of the individual employee. In addition, AI has great difficulty in incorporating principles of justice and value systems into a decision (ethical reasoning).
It is therefore advisable to equip instructing AI systems with a remonstration function and to encourage employees to make use of the function if they consider an AI instruction to be unreasonable.
Aspects of data protection law – "smart" software needs "smart" people – right of final decision also for instructions?
A major restriction on the use of AI can be found in Article 22 (1) GDPR: decisions which produce legal effects concerning an individual or similarly significantly affect him or her may not be based solely on automated data processing. This prohibition on automated decision-making is based on the idea that machines should not make decisions about people in order to protect them from "objectification" in the form of mere data collection. The decisive factor is whether the AI decision has a direct legal effect on the employee. An instruction within the meaning of section 106 German Trade, Commerce and Industry Regulation Act (GewO), on the other hand, is usually simply a specific form of the duty to work without changing the legal position of the employee. If, for example, an AI system assigns a certain route plan to the drivers of a delivery service, this is only based on individual personal data such as location and not on the fact that the driver can do the job particularly efficiently based on the skills identified in a personality profile. In this case, a final decision does not have to be made by a human being.
If the employer uses personal data for automated decision-making within the meaning of Article 22 GDPR, Article 13 (2) (f) and Article 15 (1) (h) GDPR require meaningful information about the logic involved and the scope and intended effects of such processing for the data subject.
Works council's co-determination right
The introduction of an AI system that enables algorithmic management is subject to co-determination. Pursuant to section 87 (1) no. 6 German Works Constitution Act (BetrVG), the works council has a right of co-determination in the introduction and use of technical devices designed to monitor the behaviour or performance of the employees. According to established case law, it is irrelevant whether the employer intends to or actually does carry out an assessment of behaviour and performance with the technical device; the only decisive factor is whether the device for monitoring is objectively suitable, as can be assumed in the case of an AI system.
In addition, section 90 (1) no. 3 German Works Constitution Act (BetrVG) (information and consultation rights when planning the use of AI), section 80 (3) sentence 2 German Works Constitution Act (BetrVG) (right to consult an expert when introducing and using AI) and, if applicable, section 95 (2a) German Works Constitution Act (need for approval for AI-generated selection guidelines) must be observed.
Algorithmic management and platform work
Algorithmic management is also used in the context of digital work platforms.
Against this background, the European Commission has drafted a Directive on improving platform work, which pursues three objectives in relation to algorithmic management on digital work platforms:
- Algorithms through digital work platforms are to be used more transparently.
- It should be ensured that there is human monitoring of compliance with working conditions.
- The right to challenge automated decisions should be guaranteed for both employees and genuine self-employed persons.
After the draft was adopted by the European Parliament in December 2022, it is currently the subject of trialogue between the Parliament, the Commission, and the Council of Ministers.
Exploiting the potentials of AI
AI is well-suited for use in the workplace and can help automate daily selection and deliberation processes and increase a company's efficiency. The potentials of AI use in companies come with new challenges. This makes it all the more important to define the legal framework from the outset. Only then can the use of AI be of lasting benefit.
Don't let your business fall behind on critical legal developments. Stay up to date with Law-Now, the CMS subscription service that keeps you informed on the latest legal developments. Sign up today and ensure that you never miss an important update again.