Automation in Geotechnical Engineering: AI-Enabled Foundation Design Workflow-PPT Presentation

PPT Presentation

Pdf Summary

The presentation explores how artificial intelligence could automate parts of geotechnical foundation design. It introduces neural networks for interpreting numerical data, images, and engineering charts, and generative AI for understanding natural language.<br /><br />The proposed workflow uses a network of specialized AI agents, each assigned a specific task. A site-characterisation agent extracts ground-investigation data and develops a ground model; laboratory-test agents summarize results; and a routing agent identifies the requested analysis and prepares a standardized referral, such as JSON, for a specialist agent. Separate agents could address piles, shallow foundations, retaining walls, and other engineering problems. Python and engineering software can be integrated through APIs and tools such as the Model Context Protocol, while retrieval-augmented generation (RAG) can provide access to stored project data and design knowledge.<br /><br />The central recommendation is a hybrid approach: use generative AI for tasks such as interpreting requests, mapping inputs, and drafting reports, while relying on deterministic code and rule-based engineering methods for calculations and controls. Large language models alone are generally insufficient for dependable engineering work, and programming skills can help build more capable workflows.<br /><br />AI may offer benefits including data processing, natural-language interaction, knowledge retrieval, and report generation. However, its outputs can be inconsistent, and it may hallucinate, lack sound physical intuition or genuine reasoning, and show overconfidence or limited situational awareness. The presentation therefore frames AI as a tool that must be carefully instructed, checked, and combined with engineering expertise—not as a replacement for professional judgment.

Keywords

artificial intelligence

geotechnical foundation design

neural networks

generative AI

specialized AI agents

ground model

retrieval-augmented generation

Python APIs

deterministic engineering calculations

engineering judgment