ConnectionAI is working with the University of Oxford's Department of Engineering Science in a long-term research partnership to advance the use of artificial intelligence in structural steel. The research brings together ConnectionAI's extensive structural steel knowledge and industry data with Oxford's expertise in machine learning and structural engineering.

Research aim

Accelerate the translation of AI-driven ideas into practical and accessible tools for the structural steel industry.

Research pillars

Representation

Teaching machines to understand steel. Structural steel is geometric, physical and governed by codes that do not bend. The methods that work for language and images do not transfer, so the science of how a machine perceives a structure needs to be built from first principles.

Context

Every structure is the product of constraints that live outside the drawing. Code provisions, shop capability, erection limits, fabricator preferences. Our research asks how a system learns the conditions of a specific project and a specific business, rather than producing an answer that is correct in general and wrong in practice.

Trust

Capability alone does not change an industry. Steel adopts what it can rely on, and reliance has to be earned in a domain where the consequences are physical and the liability is personal. Our research asks what conditions turn capability into reliance.

Publications

• Coming soon