Barry Law Environmental and Earth Law Journal

Dec. 1, 2024

Many people associate “artificial intelligence”, or “AI” with its portrayal in media entertainment, as human innovation leading to either catastrophic consequences or innovational benefit. This rise of AI raises an important question: Do we truly understand the relationship between AI and our world? Or perhaps a more pressing inquiry, how does artificial intelligence influence our future in the field of sustainable technology and environmental impacts? The Artificial Intelligence Environmental Impacts Act of 2024 may present a critical step toward answering these questions to uncover an understanding of the interdependent relationship between AI and resource usage.

Innovation with Data

The term “artificial intelligence” means a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. [1] “Artificial intelligence systems use machine and human-based inputs; perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate options for information or action.” [2] The use of such tools has increased as advancements continuously evolve. AI is used to gather research and formalize predictions on the environmental impact these advancements have. This data is used to assist in the development of agency guidelines and enforcements, while also allowing the government to gain empirical data to project future outcomes.

“When tackling environmental challenges, leaders depend on precise predictions to guide their decision-making processes effectively.” [3] “Researchers at Stanford have shown the Environmental Protection Agency could improve the efficiency of its Clean Water Act inspection targeting by as much as 600 percent with machine-learning algorithms” through using AI to project facilities which are not in compliance with regulations. [4] “The rapid growth of AI technologies and its data formulating capabilities offer remarkable potential for innovation but comes with environmental challenges that cannot be ignored”. [5] Although many private organizations and data collection companies have performed synergy research in evaluating the resource usage for the production and deployment of AI, more publicly accessed data is needed to derive the true effects on the environment from the advancements of AI models, specifically pertaining to the running of AI models and its support systems.

Resource Use for the Production of AI

As AI models grow in complexity, their environmental footprint expands, highlighting the need for sustainable practices in AI development and deployment. “It’s estimated that Open AI’s ChatGPT consumes 2.9 Wh per request while a typical Google search uses just 0.3 Wh and generating an image via an AI model requires the same amount of power necessary to fully charge a smartphone.” [6] Due to the emerging need to meet the demand for advanced AI as opposed to “simple AI”, larger hyperscale data center is in growing demand. “If current trends in AI capacity and adoption persist, the number of AI server units is projected to reach 1.5 million annually by 2027. Operating at full capacity, these servers would consume at least 85.4 terawatt-hours of electricity each year”. [7] In recent years there has been a surge in the numbers of hyperscale data centers. Hyperscale data centers are in wide use “globally for numerous providers and a wide variety of purposes that include artificial intelligence (AI), automation, data analytics, storage, processing and other big data computing pursuits.” [8] The International Energy Agency (IEA) projects that data centers’ electricity consumption in 2026 will be double that of 2022 — 1,000 terawatts, roughly equivalent to Japan’s current total consumption. [9]

Regulation of Hyperscale Data

The Artificial Intelligence Environmental Impacts Act of 2024, Senate Bill 3732 (2024), a legislative proposal introduced in the U.S. Senate on February 1, 2024, serves to gain research data through examining the energy consumption, pollution, and electronic waste generated throughout the lifecycle of AI models and hardware to evaluate the environmental footprint of AI technologies. “If passed, the legislation that would require the National Institutes of Standards and Technology (NIST) to develop standards for evaluating the environmental impact of AI models.” [10]

One of these standards would be to bring together a collective ranging from academia, industry, to civil society who identify methods for measuring and reporting on the environmental impacts in supporting and advancing AI. It would aim to “develop open-source software and hardware tools and provide recommendations for promoting positive environmental impacts of AI and mitigating negative ones. NIST would also be tasked with creating a voluntary and publicly available reporting system on AI models’ climate impacts.” [11]

Conclusion

AI systems rely on massive data centers for storage and computation. These centers require continuous power not only for running hardware but also for cooling systems to prevent overheating. As AI adoption grows, so does its energy consumption, raising concerns about sustainability, particularly in regions where electricity is generated from fossil fuels. While AI relies heavily on energy, its potential to transform energy systems offers a pathway to reduced environmental impact and greater efficiency. If enacted, the proposed act will facilitate expanded data collection to better understand the true environmental impact of artificial intelligence in hopes to counteract the energy usage and the positive impact artificial intelligence has on our future.

Written by Polina Mustazza, Associate Editor 2024-2025.