AI Integration and Transformation

Mission Statement/Objectives
The previous working group TG3.6 has built a strong foundation and established the "AI in AEC" conference for pioneer engineers using AI. However, the current advancements in generative AI with human-level intelligence are progressing much faster than the average pace of the bridge and structural design industry. 

Based on current research and engineering practices, we aim to establish a global and systematic approach, to prepare the entire engineering community for the transition to the new Alera.

There are generally three phases towards integrating structural design with Al:

1. Digitalization of knowledge (creating database suitable for Al model training)
2. Automation of design process (non-generative, repetitive work with fixed rules)
3. Al-assisted design (involves generative Al for decision-making support)

Key works, such as a comprehensive roadmap, a unified taxonomy, basic databases and eventually worked examples, are urgently needed to provide guidance for a timely integration with Al through the three phases mentioned above

Scope & Limitation

A group of tasks will be carried out to achieve our mission:

Task 1: Review of current status and proposition of a comprehensive roadmap of the digitalization and Al integration in bridge and structure engineering

Task 2: Digitalization of bridge and structural engineering knowledge (establishment of the database of bridge structure engineering, based on IABSE data and other public data)

Task 3 (optional): Worked example of data-driven solution for preliminary design/ complex structural analysis/ maintenance/ social-economy impact study.

Expected Project Output

From Task 1:

• Articles on the review and roadmap toward Al-integrated BSE

From Task 2:

• Guideline: Taxonomy of Bridge and Structural Engineering towards Digitalization and Al
• Database: Graph dataset of different type of bridges/structures, suitable for Vision Model
• Database: Text dataset consolidating knowledge from IABSE publications, suitable for training/tuning Large Language Model training/tuning

Start Date: 1 July 2024
Target Date of Completion: 30 June 2026

Chair
Xiaomeng Wang, Switzerland

Vice Chair
Andrey Sizikov, United Kingdom

Members
Linda Cusumano, Sweden