Abstract

AI and machine learning are key technologies in many domains and applications. While they are considered indispensable in many fields, their impact in engineering has been limited. The ambition of Scientific Machine Learning, which combines tools from both machine learning and scientific computing, is to challenge this status quo.

In this talk, we will review the opportunities and challenges of Scientific Machine Learning in an industrial context. We will discuss selected technologies that allow for faster predictions, including ML-accelerated Simulators, ML-based Model Order Reductions, and Large Language Models that foster the democratization of Computer Aided Engineering tools. We will conclude the talk by highlighting open challenges for research from an industrial perspective.