The concept of artificial intelligence (AI) has been around since the 1950s, but technology is In recent years, it has evolved at an accelerated pace. Ever since OpenAI commercialized commercial access to the model, the generated AI has been in the hands of the world's population and has been used as part of the workflow.
Many of the excitement surrounding this technology is surrounded, but its environmental impact cannot be ignored as it remains difficult to quantify.
Environmental Impacts of AI: Energy, Emissions, and Water Usage
Every time you use the model, perhaps by asking ChatGpt to summarise the email, the computer that performs the operation consumes about five times more energy than a basic web search, the researchers say.
“But every day users don't think much about it,” says Noman Bashir, lead author of the 2024 paper.The impact of generation AI on climate and sustainability“I co-written with my MIT colleagues.”
However, generative AI costs are not only limited to their use. There is a lengthy process of training the model before generating coherent results. This training is done in large data centers and requires a lot of electricity.
“The difference between the generated AI is the power density required,” says Bashir. “Essentially, it's just computing, but a generative AI training cluster can consume seven or eight times more energy than a typical computing workload.”
A data center is a temperature controlled building that houses computing infrastructure. Large companies like Amazon have one Hundreds of data centers all over the world.
The power required for training models such as GPT-3, built by Openai, is difficult to determine. Google and the University of California, Berkeley Scientists Estimated in 2021 Training this model alone consumed approximately 1,286 megawatts of electricity and produced approximately 552 tons of carbon dioxide.
However, electricity is not the overall price for this technology, as data centers use enormous amounts of water to cool electrical components during and after construction. According to one estimate, AI-related infrastructure could quickly consume six times more water than Denmark, a country of six million. A quarter of humanity There is already a lack of access to clean water and sanitation, deepening the problem of eating the already unsustainable growth resources of this generation AI technology.
As Bashir explains: “Just because this is called 'cloud computing' doesn't mean your hardware lives in the cloud. Data centers exist in our physical world and have a direct and indirect impact on biodiversity due to their use of water. ”
Related Articles: Climate disinformation is off track: what can we do? | Future challenges for generation AI | AI: The Energy Transition Paradox | Google's solution to the AI data center emissions problem? Nuclear power. | Apple, Google, Meta and Microsoft data center emissions can be 7.62 times higher than reported
Although information on the environmental impact of data centers is limited, it is impossible to predict how AI-based technologies and applications will affect the planet. AI-generated content is already flooded with information channels. If technology is used to generate False information about climate changewhich would further downplay the seriousness of this issue from the public's perspective.
So, what's it now? How to make AI more sustainable
Despite the costs, it is hoped that AI will help deal with some of the world's most pressing environmental emergency. Some examples of this include Destructive dr mapping Sand Emissions Chart Potential greenhouse gases of methane.
The United Nations Environment Programme (UNEP) is recommended Five main things To control environmental fallout from AI:
- Countries can establish standardized procedures for measuring the environmental impact of AI
- With support from UNEP, the government can develop regulations requiring businesses to disclose the direct environmental impact of AI-based products and services.
- High-tech companies can make AI algorithms more efficient and reduce energy demand
- Countries can encourage businesses to green their data centers
- Countries can incorporate AI-related policies into broader environmental regulations
“The industry is on an unsustainable path, but there are ways to encourage responsible development of generator AI to support environmental goals,” Bashir says.
there is Speculation that quantum computing will become the future of climate technology, hopefully it will be more efficient than ever computer. Computers using quantum architecture Up to 600 kWh every dayabout 1,000 times the current supercomputer. But we are still a little away from the fact that this technology is ready for industrial use.
Editor's Note: The opinions expressed here by the authors are themselves and not those of inpakter.com – In the cover photo: The original Scientific Data and Computing Centre (SDCC), previously known as the Rhic and Atlas computing facility, has grown and adapted to meet laboratory needs for decades. Covers photo credits: US Department of Energy.
