科普系列丨和ChatGPT聊天会费水?

近年来,人工智能(AI)的热潮席卷全球。在享受人工智能带来的便捷外,你知道和ChatGPT等人工智能对话还会消耗水资源吗

AI消耗水的原因是什么?科技巨头们每年会消耗多少水资源?下面和ETTBL君一起来探索一下吧~


AI的用水量惊人

The burgeoning field of generative artificial intelligence (AI), exemplified by tools like OpenAI's ChatGPT, is bringing to light a critical environmental concern: the increasing water footprint of major technology companies.

 

随着生成式人工智能(AI)领域蓬勃发展,例如以OpenAI为代表的ChatGPT日益崛起,一个重要的环保问题出现在我们的视野:科技巨头的用水量不断增加。

 

What Happened: A study led by Shaolei Ren, a researcher at the University of California, Riverside, delved into the water consumption required to operate these advanced AI models.

 

这是怎么回事?加州大学河滨分校研究员任邵磊开展了一项研究,对这些先进的人工智能模型运行所需的耗水量进行了调查。

 

Ren's research discovered that ChatGPT consumes about 500 milliliters for every 10 to 50 user prompts, which varies based on the model's deployment specifics.

 

任邵磊研究后发现,ChatGPT每回答10到50个问题时,它就会“吞下”约500毫升的水,具体耗水量因模型情况而异。

 

This data highlights the significant water usage by AI technologies, considering the vast number of users these platforms attract. CNBC reports that, if not addressed, the escalating water footprint of AI could hinder the technology's sustainable and ethical application.

 

这些平台吸引了大量用户,这一数据凸显了人工智能技术的巨大用水量。CNBC报道称,如果这一问题不加以解决,人工智能的用水量水不断增长,可能会阻碍AI的可持续发展,违背道德伦理。

 

科技巨头的用水量

Microsoft  and Google have reported significant increases in water usage, primarily attributed to the demands of AI development.

 

据报道,微软和谷歌等科技巨头的用水量大幅增加,主要原因是人工智能发展的需求。

 

For instance, Microsoft's water consumption escalated by over a third from 2021 to 2022, amounting to nearly 1.7 billion gallons.

 

例如,从2021年到2022年,微软的用水量增加了三分之一以上,将近17亿加仑。

 

Google's water usage at its data centers and offices reached 5.6 billion gallons in 2022, marking a 21% increase from the previous year. Both companies are actively working toward reducing their water footprints.

 

2022 年,谷歌数据中心和办公室的用水量达到 56 亿加仑,相较去年增加了 21%。这两家公司都在努力减少用水量。

 

"We are using tested practices to reduce the carbon footprint of workloads by large margins; together these principles can reduce the energy of training a model by up to 100x and emissions by up to 1000x," the spokesperson said, according to CNBC. 
 
据 CNBC 报道,谷歌发言人说:“经过测试,我们的方法大幅减少了工作所产生的碳足迹;这些方法一起使用,可以将训练模型的能耗减少 100 倍,排放量减少 1000 倍”。
 
AI发展面临的挑战
However, introducing AI-driven services like Microsoft's Bing Chat and Google Bard could lead to further increases in water consumption.
 
然而,引入微软必应聊天和谷歌Bard等人工智能服务可能会导致耗水量进一步增加。
 
“With AI, we're seeing the classic problem with technology in that you have efficiency gains but then you have rebound effects with more energy and more resources being used,” Joshi from the Stockholm Environment Institute told CNBC.
 
斯德哥尔摩环境研究所的乔希告诉CNBC:“在人工智能领域,我们看到了技术上的典型问题,即效率提高了,但随之而来的是产生了更多能源消耗”。
 
"And when it comes to water, we're seeing an exponential rise in water use just for supplying cooling to some of the machines that are needed, like heavy computation servers, and large-language models using larger and larger amounts of data," she added. 
 
她补充说:“说到水,我们可以看到,仅仅是冷却一些机器(如重型计算服务器和数据的大型语言模型),用水量就爆炸性增长”。
 
This growing environmental challenge poses a dilemma for Big Tech: advancing AI innovation while ensuring it does not exacerbate the strain on vital water resources.
 
这种日益严峻的环保挑战让科技巨头陷入了两难:既要推进人工智能创新,又要确保不会加剧水资源的压力。


编辑:ETTBL