Geotechnical Parrot Tales (GPT): Harnessing Large Language Models in geotechnical engineering

Krishna Kumar

2023-06-21

Abstract:The widespread adoption of large language models (LLMs), such as OpenAI's ChatGPT, could revolutionize various industries, including geotechnical engineering. However, GPT models can sometimes generate plausible-sounding but false outputs, leading to hallucinations. In this article, we discuss the importance of prompt engineering in mitigating these risks and harnessing the full potential of GPT for geotechnical applications. We explore the challenges and pitfalls associated with LLMs and highlight the role of context in ensuring accurate and valuable responses. Furthermore, we examine the development of context-specific search engines and the potential of LLMs to become a natural interface for complex tasks, such as data analysis and design. We also develop a unified interface using natural language to handle complex geotechnical engineering tasks and data analysis. By integrating GPT into geotechnical engineering workflows, professionals can streamline their work and develop sustainable and resilient infrastructure systems for the future.

Computation and Language,Geophysics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively utilize large - language models (LLMs), especially ChatGPT, in geotechnical engineering to accelerate the application of data - driven methods and computational modeling. Specifically, the paper explores the following points:

1. **Limitations of ChatGPT**: Although ChatGPT performs excellently in generating text and code, it may produce false or inaccurate outputs (i.e., "hallucinations"). The paper discusses the reasons for these hallucinations, including the influence of the model's training strategy and parameter settings (such as the temperature parameter).

2. **Importance of Prompt Engineering**: In order to reduce ChatGPT's hallucinations and fully utilize its potential, the paper emphasizes the importance of prompt engineering. Prompt engineering refers to carefully designing input prompts to guide the model to generate accurate and valuable responses. By deeply understanding ChatGPT's advantages and limitations, engineers can create effective prompts and thus obtain reliable results.

3. **Applications of ChatGPT in Geotechnical Engineering**:
- **Code Generation**: The paper shows how to use ChatGPT to generate Python code for USCS soil classification and emphasizes the importance of providing appropriate context.
- **Semantic Search Engine**: The paper proposes a semantic search engine based on vector databases and latent - space embeddings to provide correct context for LLMs, thereby avoiding hallucinations.
- **Reasoning Ability**: Through chain - of - thought prompting, the paper shows how to enable ChatGPT to have reasoning ability, such as classifying fine - grained soils and evaluating the safety of gravity retaining walls.
- **Engineering Workflow**: The paper proposes a method that combines GPT's language ability and engineering tools to solve complex geotechnical engineering problems, such as calculating the maximum bearing capacity of foundations. By introducing specific engineering tools (such as BearingCapacityTool), GPT can perform complex calculation tasks more accurately.

4. **Future Prospects**: The paper discusses the future application prospects of LLMs in geotechnical engineering, including the popularization of open - source LLMs and the development of specialized models. By integrating GPT into the geotechnical engineering workflow, professionals can complete complex tasks more efficiently and promote the development of sustainable and resilient infrastructure.

In conclusion, this paper aims to explore how to effectively utilize large - language models, especially ChatGPT, in geotechnical engineering to improve the efficiency and accuracy of data - driven methods and computational modeling.