A new bill would bring together the EPA and NIST to help Congress understand how AI is negatively or positively impacting the environment.
Lawmakers want to learn more about the environmental impacts of artificial intelligence’s potential energy consumption, with a new bill that proposes a collaborative effort between various federal agencies to assess how to mitigate environmental risks amid the rapid expansion of AI technologies.
The Artificial Intelligence Environmental Impacts Act of 2024 was introduced on Thursday by a bicameral coalition of Democratic lawmakers: Sens. Edward Markey, D-Mass; and Martin Heinrich, D-N.M.; along with Reps. Anna Eshoo, D-Calif.; and Don Beyer, D-Va.
The chief provision of the bill would require the administrator of the Environmental Protection Agency to conduct an assessment on the environmental impacts caused by AI-powered software and hardware processing, while the National Institute of Standards and Technology would be asked to convene a consortium to address these impacts.
Furthermore, the director of NIST would need to create a voluntary reporting system to register how AI technologies stand to impact the environment.
The background for this legislation stems from existing environmental problems with the tech industry. Data centers consume large sums of energy to keep hardware sufficiently cool while processing high volumes of information, further contributing to greenhouse gas emissions.
Given the established need to train AI and machine learning algorithms with large bodies of data, lawmakers are concerned this would only add to the demand for data processing centers and contribute to pollution.
“Our AI Environmental Impacts Act would set clear standards and voluntary reporting guidelines to measure AI’s impact on our environment,” Marley said in a statement. “The development of the next generation of AI tools cannot come at the expense of the health of our planet.”
The lawmakers request that the EPA’s potential report specifically include the electronic waste — or excess energy consumption — associated with the extraction of raw materials for AI systems, the exact energy and water consumption required for data center cooling, AI model design details and possible positive environmental impacts. This report would eventually be submitted to Congress.
The NIST-led consortium would feature the standards agency’s director, as well as the EPA administrator and the Secretary of the Department of Energy, among other stakeholders, who would be tasked with identifying metrics that can accurately measure and report the full range of AI’s environmental impacts.