DPS; XMANAI, EU-Projekt, News
istock / industryview

Black box AI – EU research project looks into the “inner workings” of industrial AI

News from Apr. 30, 2024

Artificial intelligence (AI) is increasingly being incorporated into industrial applications. However, AI largely acts as a black box: the calculations are often opaque to users – resulting in a lack of trust. The European research project XMANAI, involving Fraunhofer FOKUS and 14 other partners, has therefore broadened the view into AI and made its decision-making processes comprehensible.

In the XMANAI project, funded as part of the European research programme Horizon 2020, 15 research partners have developed a platform that supports the transparent use of industrial AI: Through interactive visualisations such as knowledge graphs, data diagrams and explanations written in natural language, employees can understand which data and algorithms are relevant for the calculations and how they influence the AI, even without in-depth knowledge of AI. Fraunhofer FOKUS has developed tools for data management – from importing and storing data to the (semi-)automated processing of this data – to ensure that all the data in this “explainable AI” is accurate, up-to-date and complete, thus guaranteeing the reliability of the platform. The overarching objective is to ensure low-threshold and secure utilisation of the available data.

XMANAI platform for industrial partners

In order to test the functions, visualisations and metrics of the XMANAI platform for various industrial use cases, all project partners were given access to the services via their own app and were given the opportunity to integrate them into industrial production over a period of several months. This allowed XMANAI to be tested for the specific needs and characteristics of the manufacturing industry. One of the most important findings is that the platform enables both access to AI and interoperation with on-premise environments through open, standards-based interfaces. XMANAI thus transforms “explainable AI” into the central reference point for transparent data utilisation in manufacturing processes. The results of the four-year research project are publicly accessible via the project website and will be incorporated into the improvement of industrial software applications.

The European research project XMANAI

The Horizon 2020 project XMANAI was funded by the European Commission from 01.11.2020 to 30.04.2024 as part of the Horizon 2020 research programme. The project is being led by the Italian software development company TXT e-solutions with a total of 15 European partners from business and science, including Ford and Whirlpool. The Digital Public Services (DPS) business unit at Fraunhofer FOKUS participated as a partner for real-time data management and was the scientific coordinator of the project.