Do Generative AI Tools Ensure Green Code? An Investigative Study

2024-04-16

2024 IEEE/ACM International Workshop on Responsible AI Engineering (RAIE)

AI Summary

It remains unclear whether generative AI programming tools inherently produce “green code” aligned with sustainability principles. Method: This study introduces the first Green Code Assessment Framework for Sustainable Software Engineering, integrating static code analysis, multi-tool comparative experiments (ChatGPT, Bard, GitHub Copilot), and compliance checking against established green coding guidelines—including energy-efficient algorithms and resource optimization. Contribution/Results: Empirical evaluation reveals that all three tools systematically violate core green coding principles across most scenarios; their generated code exhibits significantly higher energy consumption and lower resource efficiency than human-written sustainable implementations. The work exposes a critical sustainability gap in current AI programming assistants and delivers a reproducible methodology, empirical benchmarks, and actionable insights—thereby establishing foundational theoretical and technical support for advancing green AI-assisted software development and tool optimization.

Technology Category

• Machine Learning: Efficient ML / Green AI
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Responsible Web: Sustainability and climate impact of web technologies

Abstract

Software sustainability is emerging as a primary concern, aiming to optimize resource utilization, minimize environmental impact, and promote a greener, more resilient digital ecosystem. The sustainability or ’greenness’ of software is typically determined by the adoption of sustainable coding practices. With a maturing ecosystem around generative AI, many software developers now rely on these tools to generate code using natural language prompts. Despite their potential advantages, there is a significant lack of studies on the sustainability aspects of AI-generated code. Specifically, how environmentally friendly is the AI-generated code based upon its adoption of sustainable coding practices? In this paper, we present the results of an early investigation into the sustainability aspects of AI-generated code across three popular generative AI tools — ChatGPT, BARD, and Copilot. The results highlight the default non-green behavior of tools for generating code, across multiple rules and scenarios. It underscores the need for further in-depth investigations and effective remediation strategies.CCS CONCEPTS• Social and professional topics → Sustainability; • Computing methodologies → Natural language generation.

Problem

• Assess environmental impact of AI-generated code
• Evaluate green coding practices in AI tools
• Investigate sustainability of ChatGPT, BARD, Copilot outputs

Innovation

• Investigates AI-generated code sustainability
• Evaluates ChatGPT, BARD, Copilot greenness
• Highlights non-green default behaviors