Yves here. We have been talking about the fact that AI consumes huge amounts of energy while offering better marginal benefits than existing analytical methods. However the touts are in charge, policy makers often ignore the environmental costs and impact on grid performance.
The post below describes the weak and delayed response from the Department of Energy, as a new CNBC story (hat tip Kevin W) confirms just how bad the coming outbreak will be:
This strategy of reducing power consumption by improving computing efficiency, often referred to as “more work per watt,” is one answer to AI’s power problem. But it’s probably not enough.
A single ChatGPT query uses nearly 10 times the power of a typical Google search, according to a Goldman Sachs report. Generating an AI image can use as much energy as charging your smartphone.
This problem is not new. The 2019 estimates found one large language model training produced as much CO2 as the lifetime of five gas-powered vehicles.
Data centers building hyperscalers to meet this huge power draw are also seeing emissions increase. Google’s latest environmental report revealed that greenhouse gas emissions increased by nearly 50 percent from 2019 to 2023 in part due to data center energy use, although it also said that its data centers are 1.8 times more efficient than a typical data center. Microsoft’s output increased by about 30% from 2020 to 2024, also due in part to data centers.
And in Kansas City, where Meta is building an AI-focused data center, energy needs are so high that plans to shut down a coal-fired power plant are on hold.
There are more than 8,000 data centers around the world, with the highest concentration in the US And, thanks to AI, there will be even more by the end of the decade. The Boston Consulting Group estimates that the demand for data centers will increase by 15%-20% every year until 2030, when it is expected to include 16% of the total energy consumption in the US. That’s up from just 2.5% before OpenAI’s ChatGPT is released in 2022, and equals the power used by nearly two-thirds of all US homes.
By Haley Zaremba, a writer and journalist based in Mexico City. Originally published at OilPrice
- The rapid growth of artificial intelligence poses significant energy security risks due to its high power consumption.
- The US Department of Energy has proposed a new initiative called FASST to use AI for public benefit while addressing energy challenges and ensuring responsible AI governance.
- FASST aims to improve national security, attract skilled workers, drive scientific discovery, improve energy production, and develop AI governance technologies.
So far, the runaway growth of the Artificial Intelligence agency has proved unmanageable. As technology has taken over the tech industry like wildfire, regulators have had a lot of power to stay ahead of the spread and evolution. Questions about access and responsibility for Artificial Intelligence are being discussed, but there are few answers to go around. Then there is the problem of the energy sector and the energy growth and carbon emissions that go with it, now so important that the developed world is facing the biggest energy shortages like they haven’t seen since before the shale revolution.
“AI-powered services involve significantly more computing power – and thus electricity – than normal internet work, prompting a series of warnings about the impact of environmental technology,” the BBC reported recently. Recent research from Cornell University scientists finds that productive AI systems like ChatGPT use 33 times more energy than computers running task-specific software, and each AI-powered internet query consumes 10 times more energy than a standard search .
The global AI sector is expected to be responsible for 3.5 percent of global electricity consumption by 2030. In the United States, data centers alone could use 9 percent of electricity production by 2030, double their current levels. Already, this development is making big waves in Big Tech – earlier this month Google revealed that its carbon emissions have increased by 48 percent over the past five years.
The United States doesn’t just need more renewable growth to keep up with the insatiable demand for the tech sector, it needs more energy production, period, to avoid crippling shortages. Broad and swift action is needed on many fronts to slow the runaway train of AI adoption, but the United States also needs to keep pace with other nations’ AI spending and its own evolving national security concerns. The genie is out of the bottle, and it’s not coming back in.
“Certain strategic areas of the government’s artificial intelligence capabilities are currently lagging while foreign adversaries invest in AI,” reads a recent report from the Department of Energy (DoE). “If U.S. government leadership is not quickly established in this field, the country risks falling behind in the development of safe and reliable AI for national security, energy, and scientific discovery, thereby undermining our ability to address pressing national and global challenges.”
So the question now is not how to go back to the global adoption of AI, but how to secure new energy sources quickly, how to set strategic limits on the intensity of the sector’s growth and levels of use, and how to ensure that AI works. by being responsible and benefiting the energy sector, the nation, the community, and the world at large.
To this end, the United States Department of Energy (DoE) has proposed a new agency-wide initiative to ‘use and advance artificial intelligence for the public good’ according to a report by Axios. Just this month, the DoE released guidelines for the program, which was first discussed publicly in May this year. Frontiers in Artificial Intelligence for Science, Security and Technology (FASST) includes systematic collaboration from all 17 national laboratories of the DoE.
The program will focus on staying competitive in the AI field globally, but will also put significant resources into developing energy-efficient computer models to avoid jeopardizing the country’s energy security and climate goals in the process. The five main objectives of the program are:
- 1. Improve National Security
- 2. To attract and develop skilled workers
- 3. Harness AI for Scientific Discovery
- 4. Address Energy Challenges
- 5. Develop the technology needed for AI governance
Under the objective of “energy challenges”, the Ministry of Energy says “FASST will open up new sources of clean energy, improve energy production, and improve grid sustainability, and build a future advanced energy economy. America needs cheap energy to support economic growth and FASST can help us meet this challenge.”
While the proposed FASST program would be an important first step in the right direction for the responsible growth and use of Artificial Intelligence in the United States, it still requires congressional approval and funding for implementation. A bipartisan bill has already been introduced in the Senate.
Source link