According to research, the rise of productive applications of artificial intelligence such as chatbots and content creation systems could lead to the creation of between 1.2 and 5 million tonnes of additional e-waste (e-waste) by the end of this decade.
The study, which focuses in particular on large language models (LLM), finds that these artificial intelligence systems, thanks to their ability to interpret and produce human language, can perform diverse functions such as answering questions, writing texts or creating fulfill images.
In addition to these capabilities, the rapid development of generative AI requires more hardware infrastructure and chip updates. Researchers warn that upgrades to keep up with the growth of this technology could worsen existing e-waste problems.
The amount of e-waste can reach 2.5 million tons annually
The study says that the large computing resources required to train and operate LLMs lead to sustainability issues such as high energy consumption and carbon footprint.
The research team calculated the potential of AI to generate e-waste between 2020 and 2030 in four different scenarios. In the highest usage scenario, it is predicted that the amount of e-waste generated by AI could reach 2.5 million tons annually by 2030.
In this peak usage scenario, it is predicted that e-waste generated by AI will include 1.5 million tons of circuit boards and 500,000 tons of batteries. These batteries can contain harmful substances such as lead, mercury and chromium.
The study also notes that last year only 2.6 thousand tons of e-waste came from AI technologies, but this amount is expected to increase significantly for general e-waste.
It is estimated that the total amount of e-waste will increase by 30 percent by 2030, reaching a massive 82 million tons.
Researchers point to the importance of circular economy strategies to reduce e-waste generation.
Among these strategies, extending the life of existing infrastructure and reusing basic materials stand out.
The study highlights that using these methods can reduce the amount of e-waste created by AI by up to 86 percent.
This research was published in the journal Nature Computational Science.
Bir yanıt yazın