‘AI applications are on everyone’s lips. But the road to Safe AI systems with predictable bias is long and complex. The next generation of AI systems engineers should be trained to blend model-based systems engineering and modern AI perspectives.”
Significant advances in computational infrastructure, Big data, coupled with advanced machine learning and optimization algorithms are enabling a wide range of AI products and solutions, thus impacting all walks of society. The present wave of artificial intelligence focuses on advancing the design of context dependent, transparent, explainable systems that seamlessly integrate with humans. As engineers try to grow AI systems mimicking human intelligence by combining insights from neuroscience, cognitive sciences, applied mathematics, statistics, physics, biology and evolution, there is a fundamental need for an integrative discipline that brings together transdisciplinary perspectives involving model-based systems engineering, machine learning/AI and brain sciences to pave the way for the design, analysis and validation of complex, distributed and intelligent systems. Moreover, there is a need for future AI systems engineers to be trained in this discipline. Metaphorically speaking, “A systems engineer is like a musical maestro, who knows what the music should sound like, the look and function of a design and has the skills to lead a team in achieving the desired sound (meeting the system requirements).” At the Goethe University in Frankfurt am Main we have been developing a systems science engineering curriculum to train holistic systems thinkers, ’AI systems engineers’, with both soft-skills and technical skills required for engineering with computational platforms and tools that facilitate modern AI systems development. Specifically, these systems engineers are taught how user, modeling, implementation, and validation viewpoints can be explored for intelligent systems design and validation. This measure establishes a laboratory of artificial intelligence systems with a focus on the full AI systems engineering approach: simulation, bias avoidance, explanatory AI components, optimization of network architectures and transfer of design patterns from the brain sciences. The curriculum covers training on all aspects required for AI systems engineers: systems and software engineering, machine learning, modeling, simulation and performance analysis of AI systems, practical training, and seminar on pattern analysis and machine intelligence systems. In particular, the provision of mobility and logistics applications, autonomous driving, driver monitoring systems, video surveillance systems, bridge inspection, financial data, medicine, retail and agriculture will be of high industrial relevance. An internship offer will be created at the master level to train the AI System Engineers of tomorrow. However, the laboratory should also create added value for the specialist community in the form of courses, the opportunity to quickly build prototypes, as well as local start-ups and established companies in the field of artificial intelligence.
Summary: Future AI Systems Engineers will be provided trans-disciplinary training to become ‘holistic systems thinkers’ who understand how to translate contexts, tasks and performance requirements to AI systems that have explainability, bias-avoidance, safety certification. In addition to scientific and engineering foundations training, the training will emphasize the training of soft-skills required to be future technical leaders, managers, engineers who can exploit computational tools and infrastructure to build Integrated Human-AI systems of the future.
BMBF Program Link: https://www.softwaresysteme.pt-dlr.de/de/ki-labore.php