Quality Assurance of LLM-generated Code: Addressing Non-Functional Quality Characteristics

Xin Sun, Daniel Ståhl, Kristian Sandahl, Christoph Kessler · 2025

Read Quality Assurance of LLM-generated Code: Addressing Non-Functional Quality Characteristics by Xin Sun, Daniel Ståhl, Kristian Sandahl, Christoph Kessler (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-10271.txt.

Description

Xin Sun, Daniel Ståhl, Kristian Sandahl, and Christoph Kessler examine the quality of code that large language models generate, beyond whether it merely works. Guided by the ISO quality model, they combine a review of 109 papers, two industry workshops, and an empirical study in which three models patched real software issues. The empirical focus fell on security, maintainability, and performance efficiency, the attributes both sources flagged as critical.

Research attention concentrates on those same attributes while neglecting others, they find, whereas practitioners prize maintainability and readability and fear faster technical debt. Prompt-based attempts to optimize quality traits proved unstable in practice. The authors call the mismatch between academic focus, industry priorities, and model behavior a case for built-in quality assurance.

Catalog record

CREATORXin Sun, Daniel Ståhl, Kristian Sandahl, Christoph Kessler
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE89 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-10271

Browse the full library · About the Tower of Babel