Whether we like it, artificial intelligence (AI) is becoming part of our daily life. Online purchase to internet searches, AI is becoming an effective tool It saves people and institutions. In relation to climate changeAI proves its usefulness. At the University of Cambridge (United Kingdom), researchers use AI from climate modeling to land use planning, A converting tool to protect nature.
Researchers Oxford University They have developed the AI tool that promises to make the environmental behavior of the company’s environmental behavior. Even Google has made its advantages proud, Creating multiple AI-based tools to improve climate resistance.
Although all the possibilities of AI have a positive impact Climatic crisisThere is concern over its significant contribution to the greenhouse gas emission. The International Energy Organization (IE) shows that a new statement promotes AI Mass increase in electric demand. Data Centers are expected to be the backbone of AI systems, and in the next five years, their energy demand is doubled. By 2030, the IEA forecasts Required for electricity of data centers It will rise to about 945 teravatios Hora, ie than all the power consumption of Japan.
However, the report also indicates that the AI is capable of reducing Emuse In other areas. According to the report, if properly accepted, Carbon storage It assumes that it can compensate for the additional emissions. Fath Berol, Managing Director of IEA, says: “Artificial intelligence is one of the best stories in the world.” But so far, political leaders and markets They don’t have the tools they need To fully understand its broad consequences. “
How much energy does AI need?
AI is required Excessive energy to work and work. The enormous processing capacity required to withstand large linguistic models comes from thousands of servers placed in data centers, some of which use more energy like a small country.
Data Centers are distributed around the world, although the United States leads the list 5,381 features, about 40% of the world market. Other countries with important data centers are the United Kingdom, Germany, India, Australia, France and the Netherlands.
The energy consumption of these amenities is significant. Some IA -Regulated Data Centers They consume electricity like two million homes. In 2023, they represented around The total world consumption of electricity is 1.5%But they are expected to use the most in the coming years.
Training IA requires a lot of processor power So, a lot of electricity. A investigation published in the machine learning magazine found that exercise OpenAi’s popular satgpt model It consumed 1,287 MW of electricity, which created CO2 and 80 short -distance aircraft in Europe.
“What is the distinguishing AI that creates The density of the required power“, Says MIT’s computing and climate attack researcher Noman Bashir.” Basically, it is not only the system, but a creation AI training cluster can consume seven or eight times more energy than a common computational workload. “
Running the software uses low energy per task, but starts to accumulate quickly when millions of advice per day is sent. Electric power research agency found it, By consultation, the Saodgift uses approximately 2.9 watts. This is ten times more than the amount of energy needed for a constant search on Google.
At the beginning of 2025, SATGPT will process more than one billion per day, The number is increasing. By early 2025, about 8% of the US adults used Chadjift as the main search engine. This is still part of the number of Google users, but since it has grown from 1% in June 2024, it underlines the rapid change towards AI-based tools.
You must also take into account the changing face of AI. Current consultations are usually only for text -based interstitials. AI’s growing video, film and audio apps They have no pioneers, but they are still juicy for fuel.
“When we think The environmental impact of the creation AI. Expanding at the system level, there are constant widespread effects depending on the actions we do. “
Can I really compensate for your own emission?
According to IEA, concern I can accelerate climate change It is “exaggerated.” In spite of growth, the emissions caused by data centers will continue to be part of the emitter in the energy world, he says. In addition, the general adoption of AI can perform a series of functions more efficiently and reduce emissions in other areas. This may be due to the optimization of industrial processes, scientific research or technical innovation.
IEA calculates that the general use of the current IA-based solutions This can reduce the emission to 5% from here to 2035. This suggests that it will compensate for the increase in emission created by the need for data centers.
AdvertisementAnother Energy Intelligence Report predicts DuplicateBut provided The main cause of transforming toward clean energy. It quoted the most intelligent network administration, reduce technical costs with low carbon emissions and enhance renewable energies Benefits of AI Contributing. In addition, he confirms that the progress in the performance of the processors, the cooling technologies and the optimization of the mechanism will eventually stop the AI’s higher energy demand.
Although AIE report Look at the future of AI and its climate impact with good eyesThis results indicate that the result is not automatically. “It is important to point out that there is no trigger at present Guarantee for common adoption Of these applications of AI, “the report says.” Therefore, its additional impact, even in 2035, may be some if the necessary optimal conditions are not created. “
Ai’s ability to make a reality will require concentrated action on many fronts. Particularly, the Positive impact that can cause AI in the energy industry AI unfortunately an area that is not used today by improving networks and distribution. It also admits that investment in generating electricity with low carbon emissions is important, especially when providing Data centers that consume high energy.
Some companies are progressing in this regard. The biggest corporate buyer of renewable energy in the world of Amazon. That says that More than 90% In its operations, including its Amazon web service data centers, they are already working with renewable energy. Digital reality with over 300 data centers worldwide has selected renewable energies. Currently, 100% of its European Portfolio’s energy needs They are covered with renewable energy purchases. But Data Centers are not easy to be environmentally in business. The gap of renewable energy sources is a challenge and geographical limits, which can affect the availability of clean energy sources.
AdvertisementOur role in the future climate impacts of AI
Since most of the world’s largest data centers in the United States, it will be where it happens The greatest development of energy demand. According to the AIE report, at the end of the decade, the energy consumption of data centers is expected to be higher than all other integrated energy operations (aluminum production, concrete, chemicals).
In the present, US data centers They depend on fossil fuels, mainly natural gas. The IEA is missing that this is changing, especially with the center of the current administration in dirty fuels. During the same week, President Donald Trump signed an administrative order in which cabinet members ordered to identify regions Infrastructure provided by coal can support AI data centers.
In the state of Louisiana, there are already plans to build a large -scale gas power station to deliver New massive data center that creates the target. The AIE report presents a scene, which can only be achieved with integrated efforts and political support. Depending on the preferences of the moment, it is AI It is used to find new oil and gas reserves Finding methane leaks or improve networks.
AI should be considered in the context of the notion that its own emission “compensates”. Carbon dioxide is in the atmosphere For hundreds of years, therefore, although we find ways to reduce more emission than AI produce, It does not cancel the damage that happens in the way.
Advertisement“The general acceptance of existing AI applications can lead to Emission reduction The data is higher than the data centers, but the people who need to deal with climate change are much less than needed, ”the report concludes.
