Will AI create millions of tons of e-waste?
While there’s no denying the significant advancements AI has brought to society, from economics and education to healthcare and transportation, one less-discussed negative impact is the enormous amount of electronic waste (e-waste) this technology could produce. A recent study by the Chinese Academy of Sciences and Reichman University (Israel) suggests that AI could contribute millions of tons of e-waste over the next decade. This is a serious issue that demands our attention and specific measures to protect the environment and human health.
It is estimated that AI-generated e-waste could reach 5 million tons between 2023 and 2030. (Image: Rwanda Green Fund/Flickr)
E-waste refers to discarded electronic devices such as phones, computers, circuit boards, and batteries. Global e-waste has been increasing dramatically in recent years due to the growing demand for technology and electronic devices.
AI, especially large language models (LLMs) like ChatGPT, plays a significant role in this increase. To operate effectively, AI systems require immense computational power, forcing companies and organizations to constantly upgrade their hardware. After a certain period of use, this hardware becomes e-waste and poses environmental problems.
Scientists have outlined four different scenarios for AI production and usage from 2020 to 2030 to predict the potential amount of e-waste. In the most aggressive AI development scenario, AI-generated e-waste could reach 5 million tons between 2023 and 2030, with annual e-waste possibly reaching 2.5 million tons by the end of the decade. This is equivalent to each person on the planet discarding a smartphone.
This scenario also predicts that the additional e-waste from AI will include approximately 1.5 million tons of printed circuit boards (PCBs) and 500,000 tons of batteries containing toxic substances like lead, mercury, and chromium.
However, by considering different scenarios, researchers emphasize that AI does not necessarily have to cause such an excessive e-waste burden. The most effective strategies involve extending the lifespan and reuse of models, meaning increasing the lifespan of existing infrastructure and reusing important materials and modules in remanufacturing. Researchers suggest that implementing such circular economy strategies could reduce the e-waste burden from generative AI by up to 86%.
Although AI is contributing significantly to societal development, we cannot ignore the consequences of this technology. It is time for us to act together to develop AI sustainably, thereby minimizing its negative impact on the environment and contributing to a more sustainable world.
Source: Science Alert
