Shulou(Shulou.com)11/24 Report--

According to news on September 4, as Microsoft, Amazon, Google and Meta, the parent company of Facebook, have developed and deployed artificial intelligence, the image of big technology companies that protect the environment and focus on sustainable development is very different from the new reality.

Due to the huge power consumption of the GPU used to train large models, and the operation of the new data center requires a lot of electricity and water, the impact of artificial intelligence on the environment looks more and more like a hazard.

Amazon's vast data center in northern Virginia consumes more electricity than the entire power grid in Seattle. Google's data center consumed 5.2 billion gallons of water in 2022, an increase of 20 per cent over the previous year. The Meta artificial intelligence model Llama 2 also consumes a lot of water during its operation.

Tuyuan Pexels, as a major player in the field of artificial intelligence, these big companies are promoting how they can offset the increasing environmental burden through some environmental protection projects. Microsoft, for example, promises that its data center in Arizona does not use water for more than half of the year. Google recently announced a partnership with artificial intelligence chip giant Nvidia to replenish fresh water by 120% of the company's office and data center water consumption by 2030.

Adrianna Russell (Adrienne Russell), co-director of a research center at the University of Washington, believes that these practices may be clever marketing by the company.

"for a long time, the technology industry has been trying to make digital innovation look environmentally sustainable, but this is not the case," she says. "

To illustrate how technology companies link digital innovation to sustainable development, she lists the marketing and presentation of cloud computing and Apple products.

Now, this kind of marketing is used to show how artificial intelligence is more environmentally friendly.

In Nvidia's latest quarterly results in August, CEO Huang Renxun said that artificial intelligence-led "accelerated computing", that is, the products sold by Nvidia, are cheaper and more energy-efficient than "general computing". On the other hand, "general computing" has a relatively greater impact on the environment and higher costs.

But the opposite is true. A recent research report by Cowen, an investment bank, estimates that data centres running artificial intelligence may consume more than five times as much power as traditional facilities. The power of each GPU supplied by Nvidia is about 400W, and each artificial intelligence server has 2000 watts. Shaolei Ren, a researcher at the University of California, Riverside, who studies how modern artificial intelligence models use resources, says the power of a typical cloud computing server is 300 to 500W.

"there are always things that come with it," Russell said. "it's a lie that sustainability and digital innovation go hand in hand, such as' you can keep growing', 'everything can be massively expanded, and it's still fine'. And a technology that works for everyone."

As more enterprises want to integrate large language models into more businesses, the impact of artificial intelligence on the environment may intensify.

Russell believes that a better approach is to focus on innovations that achieve sustainable development, such as distributed networks and localized data privacy initiatives. Online communities can control data processing and network connections in a way that is less dependent on large technology companies according to their own conditions.

"if you find some examples, no matter how small, as long as they are actually designing sustainable technologies, then we can start to criticize these large technology companies that are not environmentally and socially sustainable," she said. "

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.