AI Could Contribute Millions of Tons of E-Waste, Warn Scientists

AI has rapidly advanced in replicating human activities, such as holding conversations, creating art, and producing videos. However, a study from the Chinese Academy of Sciences and Reichman University warns that AI could also contribute to environmental waste.

With the growing popularity of generative AI like ChatGPT, researchers estimate these systems could add 1.2 to 5 million metric tons of e-waste by decade’s end.

Resource Demands and Sustainability Issues of Large Language Models

This study focuses on large language models (LLMs), AI systems that generate human language by recognizing patterns in large datasets. These models are resource-intensive, requiring significant computational power, hardware, and infrastructure, raising sustainability concerns like energy use and e-waste.

While most research has focused on AI’s energy use and carbon emissions, less attention has gone to its physical materials and resulting e-waste. Led by resource management expert Peng Wang, the study estimates generative AI’s potential e-waste impact from 2020 to 2030 across different scenarios.

E-Waste Scenarios for Widespread AI Adoption

In their analysis, the researchers propose four scenarios with varying levels of AI adoption. In the most aggressive scenario of widespread AI adoption, generative AI could generate up to 5 million metric tons of e-waste from 2023 to 2030, with yearly waste reaching 2.5 million metric tons by 2030. This additional e-waste would include large amounts of printed circuit boards and batteries containing hazardous materials like lead and mercury.

In 2022, e-waste from AI-related technology amounted to only 2,600 tons. With global e-waste projected to reach 82 million tons by 2030, AI may further worsen this critical issue.

Circular Economy Solutions to Reduce AI-Related E-Waste

Peng and his team suggest that the e-waste footprint of generative AI does not have to be so high. The International Energy Agency and tech companies support circular economy strategies, like extending infrastructure lifespan and reusing materials, to manage e-waste. These circular strategies, if implemented, could cut AI-related e-waste by as much as 86%, according to the study.

Read the original article on: Science Alert

Raed more: Electricity Generated From Styrofoam Static And Wind Energy