LLMs frequently reverse their stated coding preferences when shown actual code instead of descriptions, show positional bias, and produce more polarized ratings than human experts on complexity, commenting, modularity, and readability.
InPro- ceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 4334–4353
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.HC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Subjective Code Preferences in Experts and Large Language Models
LLMs frequently reverse their stated coding preferences when shown actual code instead of descriptions, show positional bias, and produce more polarized ratings than human experts on complexity, commenting, modularity, and readability.