LLMs implement 35.3% of benchmark projects in Python despite recommending it only 10.7% of the time, mostly because the choice is automatic rather than deliberate.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LangChoiceBench: Measuring and Explaining Programming-Language Choice in LLMs
LLMs implement 35.3% of benchmark projects in Python despite recommending it only 10.7% of the time, mostly because the choice is automatic rather than deliberate.