Environmental cost (energy consumption)

Mitigation strategy

1. Mandate and incentivize the transition to renewable energy sources and carbon-aware computing for data center operations, prioritizing locations with access to low-carbon grids to decouple the substantial energy consumption of AI from greenhouse gas emissions. 2. Implement advanced algorithmic and architectural optimization techniques, such as model quantization, knowledge distillation, and the use of efficient model structures (e.g., Mixture-of-Experts or Small Language Models), to minimize computational load and energy consumption during both training and inference. 3. Establish standardized methodologies for the full life cycle analysis (LCA) of AI models, encompassing hardware manufacturing, training, and deployment, and enforce mandatory public disclosure of key environmental metrics (e.g., kWh and CO2e per model and per median query) to enable objective benchmarking and regulatory oversight.