LLMs Love Python: A Study of LLMs' Bias for Programming Languages and Libraries

2025-03-21

AI Summary

Large language models (LLMs) exhibit significant language- and library-selection biases in code generation, undermining software reliability, security, and open-source ecosystem health. Method: We conduct the first systematic evaluation of eight mainstream LLMs across benchmark tasks (e.g., HumanEval) and realistic scenarios such as new-project initialization, employing a novel multi-round prompt-engineering-based empirical framework. Results: LLMs demonstrate strong inherent preference for Python—90–97% in benchmarks and still 58% during project initialization—while systematically neglecting emerging languages and libraries. Crucially, their recommended language often contradicts the one actually generated, with an 83% inconsistency rate, exposing a fundamental flaw in decision-support capability. This work quantifies the scope and impact of language/library bias and provides both empirical grounding and methodological guidance for developing fairer, more reliable, and ecosystem-aware code-generation models.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Language and Vision

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Programming language and library choices are crucial to software reliability and security. Poor or inconsistent choices can lead to increased technical debt, security vulnerabilities, and even catastrophic failures in safety-critical systems. As Large Language Models (LLMs) play an increasing role in code generation, it is essential to understand how they make these decisions. However, little is known about their preferences when selecting programming languages and libraries for different coding tasks. To fill this gap, this study provides the first in-depth investigation into LLM preferences for programming languages and libraries used when generating code. We assess the preferences of eight diverse LLMs by prompting them to complete various coding tasks, including widely-studied benchmarks and the more practical task of generating the initial structural code for new projects (a crucial step that often determines a project's language or library choices). Our findings reveal that LLMs heavily favour Python when solving language-agnostic problems, using it in 90%-97% of cases for benchmark tasks. Even when generating initial project code where Python is not a suitable language, it remains the most-used language in 58% of instances. Moreover, LLMs contradict their own language recommendations in 83% of project initialisation tasks, raising concerns about their reliability in guiding language selection. Similar biases toward well-established libraries further create serious discoverability challenges for newer open-source projects. These results highlight the need to improve LLMs' adaptability to diverse programming contexts and to develop mechanisms for mitigating programming language and library bias.

Problem

• Investigates LLM biases in programming language selection
• Examines reliability of LLM recommendations for libraries
• Highlights need to reduce Python preference in code generation

Innovation

• Investigates LLM preferences for programming languages
• Assesses biases toward Python and established libraries
• Highlights need for mitigating LLM language bias