Do Developers Adopt Green Architectural Tactics for ML-Enabled Systems? A Mining Software Repository Study

TL;DR

The paper tackles the environmental impact of ML-enabled systems by assessing real-world adoption of green architectural tactics. It builds on the Järvenpää catalog of 30 tactics and uses a novel LLM-based mining mechanism to detect tactic usage in 168 GitHub ML projects. Results show widespread adoption of several tactics (e.g., built-in library usage, memory-conscious design), but weaker uptake for energy-constrained and graph-substitution tactics, with larger projects showing more resistance. Contributions include an empirical analysis of tactic adoption, development of an LLM-driven extraction tool, and a public replication package to enable further study. This work informs developers and toolmakers about practical green AI strategies and sets the stage for automation and broader adoption in ML systems.

Abstract

As machine learning (ML) and artificial intelligence (AI) technologies become more widespread, concerns about their environmental impact are increasing due to the resource-intensive nature of training and inference processes. Green AI advocates for reducing computational demands while still maintaining accuracy. Although various strategies for creating sustainable ML systems have been identified, their real-world implementation is still underexplored. This paper addresses this gap by studying 168 open-source ML projects on GitHub. It employs a novel large language model (LLM)-based mining mechanism to identify and analyze green strategies. The findings reveal the adoption of established tactics that offer significant environmental benefits. This provides practical insights for developers and paves the way for future automation of sustainable practices in ML systems.

Figures (2)

• Figure 1: Overview of the research process.
• Figure 2: Frequency of Green Tactics in ML Projects.