Between 1,200 and 5,000 million tons of electronic waste, the equivalent of 20,000 million smartphones, between 2020 and 2030. AI and its development pollute more than we think. Researchers from the Chinese Academy of Sciences and the University of Cambridge show the extent of this in this study published in the scientific journal Nature Computational Science.
Tech, soon one of the most polluting sectors?
In this context, we talk about e-waste or waste electrical and electronic equipment (WEEE). These include most appliances, screens and phones. But the WEEE that is at the origin of this explosion of pollution are electronic components such as cards or chips found in our computers.
These are wastes that pollute due to the presence of certain especially harmful products or components. According to the Ministry of Ecology, it contains mercury, PCBs and certain greenhouse gases.
Why does the technology sector generate so much waste? In the world of technology, innovation and competition are king. As a result, product obsolescence becomes a problem that all companies quickly face. An electronic card is quickly replaced by a more powerful model. And as the development of generative AI requires more and more resources, this further accelerates this race for the most efficient components.
The equivalent of 10 billion smartphones in 2030
According to the study, in 2023, generative AI generated almost 2,600 tons of WEEE. In 2030, if no measures are taken, the worst scenario predicted by researchers will multiply this amount of waste by 1,000. It is estimated at 2.5 million tonnes of e-waste, representing 10 billion discarded smartphones (250g each) in a single year.
This exponential increase can be explained by a large-scale democratization of artificial intelligence in the coming years. Researchers have already observed a sharp increase in WEEE between 2020 and 2023, during which ChatGPT and other generative AI began to emerge.
Including all WEEE (which includes, for example, household appliances), UNITAR (United Nations Institute for Training and Research) forecasts 82 million tonnes of waste by 2030 globally.
A possible mitigation
Researchers foresee an alternative, more optimistic scenario. It would reduce WEEE between 16 and 86%. Several avenues remain to be explored, including a downcycling process: electronic components could be recycled into lower-value objects.
Researchers also tend to favor a few recent, powerful components rather than multiple less potent components. Finally, they encourage the optimization of software so that it requires fewer resources and does not accelerate the obsolescence of electronic components.
“Implementing circular economy strategies across the generative AI value chain could reduce the production of e-waste. This highlights the importance of proactive e-waste management in the face of evolving intelligence technologies.” generative artificial,” the researchers indicate.
Source: BFM TV