LOC Talk Dr. Ekaterina Petrova - Neuro-Symbolic AI in AEC

The Dartmouth Summer of 1959 envisioned Artificial Intelligence (AI) as a “program which can solve intellectual problems as well as or better than a human being”. More than six decades later, that vision remains. The strive towards intelligent machines continues in full force and is stronger than ever. In the era of digitalisation and an increasingly virtual world revolving around technological advancements and endowing machines with human-like intelligence, built assets are among the few tangible things left. Even though the built environment is still challenged by time-consuming, inefficient, and costly processes associated with capturing, identifying, analysing, accessing, reusing, and sustainably managing data, the potential of Building Information Modelling (BIM), Digital Twins and AI are indisputable.

This talk will discuss the untapped potential of neuro-symbolic AI for AEC. That entails combining symbolic (knowledge representation, automated reasoning) and subsymbolic (Machine Learning, Natural Language Processing) approaches as means for enabling a better performing built environment by addressing several shortcomings of the two AI approaches when used separately. To demonstrate the reconciliation between symbolic and subsymbolic AI for AEC, the talk will highlight recent implementations and ongoing research conducted at Eindhoven University of Technology in the context of smart building operation, circular construction, material performance assessment and advanced compliance checking.