According to Lufthansa Industry Solutions (LHIND), the EU AI Act is causing uncertainty across all industries. Even in technical fleet management, the use of artificial intelligence (AI) is therefore still lagging behind the possibilities. In order to offer companies a high level of legal certainty, LHIND has had its AI Maintenance Assistant assessed by TÜV SÜD and Calvin Risk as the first AI-supported maintenance application. This should enable effective risk management in accordance with EU law without a great deal of effort.

"The AI Regulation adopted by the EU Parliament in March is intended to create transparency and facilitate the use of AI. However, the AI Act still tends to make companies reluctant to develop smart and individualized assistants," says Michael Koch, Director Artificial Intelligence at LHIND.

Michael Koch, Director Artificial Intelligence at LHIND

© LHIND

TÜV SÜD contributed its self-developed AI quality framework, which is composed of various norms and ISO standards for AI systems and specifications of the European Committee for Standardization (CEN). The Evidence Management Tool (QLAIM) developed by Swiss technology company Calvin Risk in collaboration with TÜV SÜD is the operational software system for conducting such a digital audit and works within the framework defined by TÜV SÜD.

"TÜV SÜD's AI quality framework combines classic quality management approaches with specific regulatory requirements for AI and can also be easily implemented using the compliance tool from Calvin Risk and TÜV SÜD. In addition, the risk profile of the use case and measures to minimize risk are also taken into account. With this combination and the systematic approach, a high level of quality and legal certainty for AI applications is already possible today," says LHIND expert Koch.

Free from distortions and prejudices

According to Philippe Coution, Lead AI Quality at TÜV SÜD, the highest demands are placed on AI systems. He explains: "In addition to the safety and reliability of the systems, the results of automated data collection and processing must be free from bias and prejudice. In addition, the functioning of AI systems must be transparent and the results must be accurate and comprehensible. Social requirements are also becoming increasingly relevant and are taken into account in the TÜV SÜD AI Quality Framework. The complex requirements call for a systematic approach to quality and risk management as well as regular comparison with the relevant standards and dynamic regulation in order to maintain market suitability."

Precise maintenance technology, troubleshooting and quality assurance

When developing intelligent assistants, LHIND now not only works with text, but also with image data and videos. "The AI Maintenance Assistant is revolutionizing technical fleet management by supporting maintenance technicians, troubleshooting and quality assurance precisely and quickly," says LHIND expert Koch. Text documents as well as images and videos are used. The application - built on a secure, cloud-based system - streamlines decision-making, increases efficiency and reduces costs.