Breaking Exponential Complexity in Design & Manufacturing of Engineering Systems

The complexity of engineering challenges of our time is growing at exponential rates, while our solutions are at best improving linearly. This complexity is surpassing our ability to understand and reason about the artifacts of our creation, notwithstanding the enormous progress in artificial intelligence (AI) and high-performance computing (HPC). Computation power and affordability are rapidly improving, however, Moore’s law has slowed down and alternative computing paradigms are too nascent. No matter how powerful, HPC cannot break exponential complexity by linear speed-up. AI systems, on the other hand, are breaking records in predicting the behavior of complex systems, but lack the ability to help us understand them or question the implicit modeling assumptions. The role of human experts in developing more effective abstractions is undeniable. To help with this process, we need scalable mathematical machinery for compositional modeling that enable reasoning on complex models with a logarithmic cognitive overhead. We need a paradigm shift from imperative and point-solutions to declarative and property-based decision making, from automation to human-AI teaming, and from data-driven interpolation to hybrid (models and data) extrapolation.