PreprintCC BY 4.0arXiv
BIRDS: Characterizing and Understanding Biodiversity Impact of Large Language Model Serving
Abstract
Large language model (LLM) serving creates environmental impacts beyond carbon and water, including ecosystem damage through biodiversity-related pathways. We present BIRDS, a framework for Biodiversity Impact of Request-Driven LLM Serving. BIRDS defines request-level functional units, quantifies operational and embodied biodiversity impact, and introduces Quality-Normalized Biodiversity Impact (QNBI) to jointly analyze ecological impact and response quality. Across diverse workloads, models, GPUs, and regions, BIRDS reveals that biodiversity impact accumulates at scale and exposes quality-aware serving tradeoffs. The code is available at https://github.com/TianyaoShi/BIRDS.
Related
- Understanding the Earth system through its biodiversity: Why and how
- Biotic interactions biogeography: A framework for understanding how species interactions shape biodiversity patterns across scales
- Uneven research attention to biodiversity loss revealed by large language models
- BioCosmos: computer vision and large language models for large-scale image-based biodiversity data and trait exploration
- Gaps and biases in the ecological evidence underlying global biodiversity impact assessments
BibTeX
@misc{shi2026birds,
title = {{BIRDS: Characterizing and Understanding Biodiversity Impact of Large Language Model Serving}},
author = {Tianyao Shi and Yi Ding},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2605.27480},
eprint = {2605.27480},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.27480},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - BIRDS: Characterizing and Understanding Biodiversity Impact of Large Language Model Serving
AU - Tianyao Shi
AU - Yi Ding
PY - 2026
DA - 2026/08/29
PB - arXiv
DO - 10.48550/arxiv.2605.27480
UR - https://arxiv.org/abs/2605.27480
AB - Large language model (LLM) serving creates environmental impacts beyond carbon and water, including ecosystem damage through biodiversity-related pathways. We present BIRDS, a framework for Biodiversity Impact of Request-Driven LLM Serving. BIRDS defines request-level functional units, quantifies operational and embodied biodiversity impact, and introduces Quality-Normalized Biodiversity Impact (QNBI) to jointly analyze ecological impact and response quality. Across diverse workloads, models, GPUs, and regions, BIRDS reveals that biodiversity impact accumulates at scale and exposes quality-aware serving tradeoffs. The code is available at https://github.com/TianyaoShi/BIRDS.
ER -