Use of Large Language Models (LLM) as a Structural Engineering Design Assistant

This is a self-funded project - please only apply if you have funding to undertake this project

Large Language Models such as Chat GPT3 represent a significant development in artificial intelligence and its impact on society. Trained on a huge amount of data and able to generate useful responses to complex and multi-layered questions, it is potentially the most useful tool available for handling structural engineering tasks. However, its results are not always completely accurate and sometimes completely incorrect, though delivered with authority.

This project will develop a computational interface to a leading LLM (such as the latest Chat GPT) through its API to assess the capability of the model to act as a structural engineering design assistant. The research will aim to identify the following:

1. The limits of the model's understanding of structural engineering concepts through its ability to answer complex technical questions correctly.
2. Its ability for developing code for automating design calculations to specified standards such as Eurocodes, ASCE, API, etc. based on text or voice inputs. This will be for routine design scenarios (such as steel and concrete beam design) through to complex ones (such as buckling of plate girders, non-linear M-phi interaction, calculation of eigenvalues (mode shapes)) and the ability for sequences of multi-step calculations to be developed to assess multiple stages of a design.
3. Interpretability of code outputs and their translation into human-readable Latex-based calculation reports for verification purposes.
4. Its accuracy in writing technical reports that summarise calculation outputs as well as provide the context of the design and include relevant images to demonstrate concepts.

A wide range of automated tests such as unit testing, integration testing, functional testing and acceptance testing will need to be developed to determine the LLM's ability as a design assistant.

The output of the research will be a computational interface to an LLM that can operate as a structural engineering design assistant to perform complex design calculations and write technical reports.

This project will require familiarity with programming languages such as Java, C++, Python or similar as well as an understanding of structural engineering design.

The successful applicant will join the Centre for AI: Social and Digital Innovation, which has more than 15 PhD researchers working in the field of computer vision, robotics, drones and includes a newly fitted artificial intelligence and robotics lab.

Brunel offers a number of funding options to research students that help cover the cost of their tuition fees, contribute to living expenses or both. See more information here: https://www.brunel.ac.uk/research/Research-degrees/Research-degree-funding. The UK Government is also offering Doctoral Student Loans for eligible students, and there is some funding available through the Research Councils. Many of our international students benefit from funding provided by their governments or employers. Brunel alumni enjoy tuition fee discounts of 15%.