The development of generative artificial intelligence and, in particular, large language modelscould generate between 1.2 and 5 million tons of electronic waste accumulated between 2020 and 2030, according to a study published in Nature Computational Science. The analysis calculates the mass of garbage generated by hardware elements, such as processing units, storage units and power systems. The study was led by Peng Wang, from the Chinese Academy of Sciences; Asaf Tzajor, from Reichman University, Herzliya, in Israel and Wei-Qiang Chen, also from the Chinese Academy of Sciences.
The conclusions, said Shaolei Ren, associate professor of Electrical and Computer Engineering at the University of California at Riverside, “are supported by the best data available in the public sphere and in scientifically valid methods».
Although many studies have estimated and highlighted the global magnitude of e-waste, Ren said, this study specifically examines e-waste produced by generative AI, one of the fastest growing applications.
The study uses Nvidia’s DGX H100 server as a reference to estimate the mass of garbage coming from next-generation servers. “Although predicting future hardware development is difficult, I consider the document’s forecast to be a reasonable indicator of future developments.” electronic waste that AI will probably generate.
E-waste is a critical, yet often overlooked, issue when considering the future societal impact of AI. Therefore, with this article, Ren concludes: «Attention is drawn to the electronic waste generated by AI and I believe it will invite a deeper debate. And what is more important, highlights the crucial role of a circular economyr to achieve truly sustainable AI.”
Report a bug
#produce #tons #electronic #waste
