ExamAI – KI Testing & AI Auditing
It is already foreseeable today that the use of artificial intelligence (AI) will have a significant impact in the workplace and will continue to do so in the future. The requirements raised in the socio-political debate on "AI-based systems" - such as transparency, traceability, fairness, equal treatment, liability, reliability or data protection - are unclear in terms of their legal and technical design and sometimes conflict with the interests of the companies developing them.
Objectives & Approach
The project aims to make the operation of AI systems traceable, measurable and thus certifiable. Concrete concepts for the auditing and certification of AI applications are to be developed. For this purpose, the technical and legal requirements to be considered when using AI systems will be determined. The investigation is based on two typical fields of application for AI systems.
AI in Production Processes
One application scenario that concerns the use of AI systems in production processes, is for example in the form of collaborating robots ("cobots") and driverless transport systems (AGVs). The focus of this scenario is the safe design of the human-machine interface to create a safe working environment. By developing test and certification procedures, risks can be minimized and trust and legal certainty for the use of AI can be created.
AI in Personnel Decisions
The second application scenario deals with the risks of using AI-based systems in human resource management. Here, certifications should help to ensure that AI systems work without discrimination and within the framework of the legal requirements. Furthermore, it ensures that decision proposals are comprehensible in order to prevent possible liability risks
Funded by the federal ministry of labour and social affairs.
Project Manager |
Chair of Prof. Dr. Georg Borges for Civil Law, Legal Informatics, German and International Business Law and Legal Theory |
Contact |
Elisabeth Friedel (Mail) |
Research Associates |
Robert Hoffmann, Dr. Andreas Sesing (Mail) |
Project partner |
Gesellschaft für Informatik e.V. |
Project duration |
March 2020 – November 2021 |
Project description |
Here you can download the project profile as pdf (in German only). |