BIRDS: Characterizing and Understanding Biodiversity Impact of Large Language Model Serving

arXiv:2605.27480 · q-bio.OT, cs.AI, cs.CY · Submitted 2026-05-26 · Read on arXiv

q-bio.OT, cs.AI, cs.CY

Submitted: 2026-05-26

Updated: 2026-08-29

Terminology

• Benchmark — 17× in this paper · explained in 2 episodes
• Trace — 9× in this paper · explained in 2 episodes
• Water Consumption — 8× in this paper
• Accuracy — 6× in this paper · explained in 1 episode
• RAG — 6× in this paper · explained in 3 episodes
• Global Warming — 5× in this paper
• Quality-Normalized Biodiversity Impact — 4× in this paper
• Terrestrial Acidification — 4× in this paper
• Core — 3× in this paper · explained in 2 episodes
• KV cache — 2× in this paper · explained in 3 episodes
• Large Language Model (LLM) — 2× in this paper · explained in 3 episodes
• Thinking mode — 2× in this paper · explained in 2 episodes
• United States Environmental Protection Agency — 2× in this paper
• Agentic LLM Systems — 1× in this paper · explained in 1 episode
• Agentic serving — 1× in this paper · explained in 1 episode
• Aggregation — 1× in this paper · explained in 1 episode
• Benchmarking — 1× in this paper · explained in 3 episodes
• European Commission Joint Research Centre — 1× in this paper
• Foundation Model — 1× in this paper · explained in 3 episodes
• Foundation Models — 1× in this paper · explained in 7 episodes
• Freshwater Eutrophication — 1× in this paper
• Freshwater Toxicity — 1× in this paper
• Generative AI — 1× in this paper · explained in 2 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Marine Eutrophication — 1× in this paper
• Marine Toxicity — 1× in this paper
• Mixture-of-Experts (MoE) — 1× in this paper · explained in 7 episodes
• Reasoning Traces — 1× in this paper · explained in 3 episodes
• Terrestrial Toxicity — 1× in this paper

Sources

• LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding
• More than Carbon: Cradle-to-Grave environmental impacts of GenAI training on the Nvidia A100 GPU
• The Llama 3 Herd of Models
• SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines
• Llama 2: Open Foundation and Fine-Tuned Chat Models
• gpt-oss-120b & gpt-oss-20b Model Card
• Qwen3 Technical Report

Related papers

• A mechanistic interpretation of mutation risk across biological scales
• Compositional proofreading through critical self-tuning
• 2D reaction-diffusion model-based biopsy simulation for dynamic tumor growth parameter estimation
• Flash-Radiomics: A Scalable Hybrid CPU-CUDA Engine for Standardized Scalar Radiomics and Accelerated Spatial Mapping
• Recovery Rates Are Not Comparable Across Transcription Factors: Chance Correction for Attribution Evaluation
• Geochemical Hazard Assessment of Martian Regolith for Future Human Exploration