aiqbench analysis
This research identifies a critical vulnerability where LLMs generate unit tests that validate buggy code rather than detecting it, quantifying this "misguidance effect" through new metrics and internal model analysis. The authors propose replacing direct code prompts with specification-based docstrings to effectively mitigate the issue while improving test generation pipelines for both flawed and correct software. This work advances automated testing reliability by addressing a fundamental limitation in current LLM-driven development workflows.
Evaluating and Mitigating the Misguidance Effect of Buggy Code in LLM-Generated Unit Tests
arXiv:2607.22883v1 Announce Type: cross Abstract: While Large Language Models (LLMs) show great promise for automating unit test generation, recent studies suggest that the quality of generated tests can be negatively impacted when models are prompted with buggy code. This paper presents a new metric to quantitatively measure the "misguidance effect," a phenomenon where buggy code steers LLMs toward generating tests that validate its erroneous behavior rather than expose it. Our analysis reveals that prompting LLMs with buggy code has a severe, twofold impact: it significantly increases "misguided tests" that assert incorrect behavior while simultaneously suppressing the generation of effective, bug-finding tests. We further corroborate this effect from a model-internal perspective, showing that buggy code skews LLMs' preference toward tests that assert the same erroneous behavior. To counter this, we introduce and validate a specification-based unit test generation paradigm that replaces the code under test in the prompt with an LLM-generated specification docstring. Our results show that this paradigm effectively reduces misguided tests while substantially increasing effective tests, improves multi-round, feedback-driven test generation pipelines, and remains applicable to both buggy and bug-free code. Overall, these results suggest that specification-based prompting is a promising strategy for mitigating misguidance from buggy code in LLM-generated unit tests.
Abstract:While Large Language Models (LLMs) show great promise for automating unit test generation, recent studies suggest that the quality of generated tests can be negatively impacted when models are prompted with buggy code. This paper presents a new metric to quantitatively measure the "misguidance effect," a phenomenon where buggy code steers LLMs toward generating tests that validate its erroneous behavior rather than expose it. Our analysis reveals that prompting LLMs with buggy code has a severe, twofold impact: it significantly increases "misguided tests" that assert incorrect behavior while simultaneously suppressing the generation of effective, bug-finding tests. We further corroborate this effect from a model-internal perspective, showing that buggy code skews LLMs' preference toward tests that assert the same erroneous behavior. To counter this, we introduce and validate a specification-based unit test generation paradigm that replaces the code under test in the prompt with an LLM-generated specification docstring. Our results show that this paradigm effectively reduces misguided tests while substantially increasing effective tests, improves multi-round, feedback-driven test generation pipelines, and remains applicable to both buggy and bug-free code. Overall, these results suggest that specification-based prompting is a promising strategy for mitigating misguidance from buggy code in LLM-generated unit tests.
24 pages, 7 figures, 12 tables. Accepted at ISSTA 2026; to appear in Proceedings of the ACM on Software Engineering (PACMSE), Vol. 3, No. ISSTA, Article ISSTA113
Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2607.22883 [cs.SE]
(or arXiv:2607.22883v1 [cs.SE] for this version)
https://doi.org/10.48550/arXiv.2607.22883 arXiv-issued DOI via DataCite (pending registration)
https://doi.org/10.1145/3832204 https://doi.org/10.1145/3832204 DOI(s) linking to related resources
Full text under the publisher's open licence · 256 words
