Authors: Ma, Wei and Liu, Shangqing and Lin, Zhihao and Wang, Wenhan and Hu, Qiang and Liu, Ye and Zhang, Cen and Nie, Liming and Li, Li and Liu, Yang
Abstract:
Large language models~(LLMs) demonstrate significant potential to revolutionize software engineering (SE) by exhibiting outstanding performance in SE tasks such as code and document generation. However, the high reliability and risk control requirements in software engineering raise concerns about the lack of interpretability of LLMs. To address this concern, we conducted a study to evaluate the capabilities of LLMs and their limitations for code analysis in SE. We break down the abilities needed for artificial intelligence~(AI) models to address SE tasks related to code analysis into three categories: 1) syntax understanding, 2) static behavior understanding, and 3) dynamic behavior understanding. Our investigation focused on the ability of LLMs to comprehend code syntax and semantic structures, which include abstract syntax trees (AST), control flow graphs (CFG), and call graphs (CG). We employed four state-of-the-art foundational models, GPT4, GPT3.5, StarCoder and CodeLlama-13b-instruct. We assessed the performance of LLMs on cross-language tasks involving C, Java, Python, and Solidity. Our findings revealed that while LLMs have a talent for understanding code syntax, they struggle with comprehending code semantics, particularly dynamic semantics. We conclude that LLMs possess capabilities similar to an Abstract Syntax Tree (AST) parser, demonstrating initial competencies in static code analysis. Furthermore, our study highlights that LLMs are susceptible to hallucinations when interpreting code semantic structures and fabricating nonexistent facts. These results indicate the need to explore methods to verify the correctness of LLM output to ensure its dependability in SE. More importantly, our study provides an initial answer to why the codes generated by LLM are usually syntax-correct but vulnerable.
Link: Read Paper
Labels: static analysis, data-flow analysis, call graph analysis, data-flow analysis, code model, code model training, source code model, empirical study