The Artificial Intelligence Environmental Impacts Act of 2024: What You Need to Know

When used responsibly and with the appropriate safeguards, AI can have incredible transformational potential for organizations. However, these powerful models can have a significant environmental impact due to the amount of energy required to train and deploy models, as well as the huge amounts of water needed to cool the servers. This has led to increasing concerns about the sustainability of AI and the environmental impact of the technology, particularly in light of hugely complex generative AI models.

In response to these concerns, on 1 February 2024, the Federal Artificial Intelligence Environmental Impacts Act of 2024 was introduced by Senator Markey and co-sponsored by several senators. The bill addresses the need to assess and mitigate the environmental impacts of AI technologies and emphasizes the importance of understanding both the positive and negative effects of AI on the environment, aiming to promote transparency and accountability in the development and use of AI systems. In this blog post, we outline the key things you need to know about the AI Environmental Impacts Act of 2024.

How does the AI Environmental Impacts Act define AI?

The Environmental Impacts Act defines AI, under section 5002 of the National Artificial Intelligence Initiative Act of 2020, as:

“a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. Artificial intelligence systems use machine and human-based inputs to: a) perceive real and virtual environments; b) abstract such perceptions into models through analysis in an automated manner; and c) use model inference to formulate options for information or action.”

This definition shares some similarities with many other definitions of AI in terms of the role of humans, being a machine-based system, and having a range of outputs.

The bill also distinguishes between AI and an AI model, where a model is defined as:

“a component of an information system that implements artificial intelligence technology and uses computational, statistical, or machine-learning techniques to produce outputs from a given set of inputs”.

Moreover, the Act defines an AI system as “any data system, software, hardware, application, tool, or utility that operates in whole or in part using artificial intelligence.”

What are the main requirements of the AI Environmental Impacts Act?

Although a relatively short bill, the AI Environmental Impacts Act has four key provisions, all of which the National Institute of Standards and Technology (NIST) is responsible for:

1. A study on the environmental impacts of AI and submission of a report to Congress
2. The establishment of a consortium of stakeholders to investigate the environmental impacts of AI
3. The establishment of a system for voluntary reporting by entities on the environmental impacts of artificial intelligence.
4. A report to Congress on findings surrounding the environmental impact of AI

Overall, the Artificial Intelligence Environmental Impacts Act promotes transparency, accountability, and understanding of the environmental footprint of AI to address environmental concerns surrounding AI technologies. Through standardized measurement, collaborative studies, and voluntary reporting, it aims to ensure that AI development aligns with environmental sustainability goals.

What are the requirements of the study on the environmental impacts of AI?

One of the core provisions of the Bill is the requirement for the Administrator of the Environmental Protection Agency (EPA) to conduct a comprehensive study on the environmental impacts of AI. This study will cover various aspects, including energy consumption, pollution, and potential local-scale impacts.

The study requirements include:

• Examining the energy consumption and pollution associated with the entire lifecycle of AI models and hardware.
• Investigating the energy and water consumption for cooling data centers used in AI activities.
• Analyzing how design, development, deployment choices, and hardware types impact environmental impacts.
• - Identifying potential local-scale environmental impacts such as grid stress and water withdrawals.
• Assessing both positive (e.g., energy efficiency optimization) and negative (e.g., rebound effects) environmental impacts of AI applications.
• Addressing disparate impacts in negative environmental impacts and considering other environmental impacts as determined by the EPA Administrator.
• Incorporating the results of an updated data center study as required by the Energy Independence and Security Act of 2007.

Additionally, the Administrator must solicit and consider public comments during the study process.

What is the role of the environmental impact Consortium?

Section 5 of the Bill establishes the Artificial Intelligence Environmental Impacts Consortium, overseen by the Director of the NIST in consultation with relevant authorities, including the Administrator of the EPA and the Secretary of Energy, among others. This consortium will consist of stakeholders from academia, civil society, and industry and will focus on identifying future measurements, methodologies, standards, and other necessary elements to comprehensively assess and report the environmental impacts of AI.

The consortium's goals include promoting consistent and comparable reporting on the environmental impacts of AI throughout its lifecycle, fostering the development or compilation of open-source tools for measuring these impacts, and providing recommendations for mitigating negative impacts while promoting positive ones associated with AI technologies. Ultimately, the consortium aims to enhance understanding and management of the environmental implications of AI advancements.

What form must the voluntary Reporting System for AI environmental impacts take?

The Artificial Intelligence Environmental Impacts Act establishes a voluntary reporting system for the environmental impacts of AI, where the Director of the NIST will lead the development of guidelines for voluntary reporting entities on how to participate in this system. These guidelines will cover various environmental impacts such as energy consumption, water consumption, pollution, and electronic waste throughout the lifecycle of AI models and hardware. Before finalizing the guidelines, there will also be an opportunity for public comments, and submissions to the voluntary reporting system will be made available to the public as much as possible.

The AI Environmental Impacts Act also mandates a joint report to Congress, to be submitted within four years of the Act's enactment. This report will include the main findings of the consortium, and recommendations for legislative or administrative actions to mitigate negative environmental impacts and promote positive ones related to AI.

Towards greener AI

In conclusion, the Artificial Intelligence Environmental Impacts Act of 2024 represents a proactive step toward addressing the environmental challenges posed by AI technologies. By conducting a comprehensive study, convening a consortium, and establishing a voluntary reporting system, the Bill aims to promote transparency, accountability, and responsible innovation in the AI sector.

As AI continues to shape our future, it is crucial to consider its broader societal and environmental implications. By taking concrete steps to understand and mitigate its environmental impacts, AI can be made more sustainable and aligned with goals for a greener and more resilient future.

Schedule a demo with our experts to find out how Holistic AI’s Tracker can help you navigate the ever-changing AI legislative landscape.