This is Hamamoto from TIMEWELL
This is Hamamoto from TIMEWELL.
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
AI Power Consumption Could Surpass Small Countries: Study Warns
AI Power Consumption Could Surpass Small Countries: Study Warns
In 2021, AI accounted for 10%–15% of Google's energy consumption
A new study published in Joule warns that the widespread use of artificial intelligence (AI) could cause a significant increase in power consumption.
As generative AI technologies that rely on powerful servers expand in use, energy demand could increase.
For Example, While AI Accounted for Only 10%–15%
For example, while Google noted that AI accounted for only 10%–15% of the company's total power consumption in 2021, it has been pointed out that if this expands, it could lead to energy consumption equivalent to a small country like Ireland.
Experts say creative ways to manage AI-related energy consumption must be found, and propose leveraging alternative energy sources such as flare gas and biogas from landfills.
While this study also references extreme and unlikely scenarios, it emphasizes the importance of managing expectations and potential energy impacts as AI technology continues to evolve.
By Michael Lee (Fox News)
Published October 12, 2023, 2:00 AM (EDT)
Published October 12, 2023, 2:00 AM (EDT)
A new study warns that artificial intelligence technology could cause a significant increase in power consumption. The paper, published in the journal Joule, describes the future energy consumption of AI systems in detail, noting that generative AI technologies rely on powerful servers and that increasing usage could lead to a surge in energy demand.
The paper's author cites tech giant Google as an example, noting that AI accounted for only 10%–15% of the company's total power consumption in 2021. However, they warn that if AI technology continues to expand, Google's energy consumption could reach a scale comparable to a small country.
According to the paper, in a worst-case scenario, Google's AI alone could consume electricity comparable to a country like Ireland (29.3 TWh annually), which would be a substantial increase compared to historical AI-related energy consumption.
The Author Warns That Such Examples
The author warns that such examples "assume full-scale AI adoption with current hardware and software and are unlikely to materialize rapidly."
Christopher Alexander, Chief Analytics Officer at Pioneer Development Group, said the demand resembles the birth of Bitcoin mining, and argued that developers must be creative in how they use resources. "AI is very similar to Bitcoin mining. In both cases, processing power is used at very high intensity to solve problems. You can't reduce the energy consumption, but you can mitigate it," Alexander said. "For example, there are alternative energy sources such as natural gas from oil wells that would be flared rather than used, and biogas from landfills is one example."
Phil Siegel, Founder of the Center for Advanced
Phil Siegel, founder of the Center for Advanced Preparedness and Threat Response Simulation (CAPTRS), said similar concerns are characteristic of growing technologies, but argued that improvements could be made to help improve energy consumption efficiency. "Multiplayer gaming, social media, and cryptocurrency all went through these phases. In the early stages, the technology tends not to be efficient because chips and algorithms have not been optimized. People extrapolate these inefficiencies at scale. The bad news is energy consumption will increase somewhat. The good news is that as new uses scale, chips improve, algorithms get better, the technology becomes more creative, and ultimately reduces energy consumption well below panic levels."
The paper acknowledges that some scenarios are extreme and unlikely cases, while emphasizing the importance of suppressing "overly optimistic and overly pessimistic expectations" about the future, and stating that it is probably overly optimistic to expect hardware and software efficiency improvements to fully offset long-term changes in AI-related power consumption.
The Paper's Conclusion States
The paper's conclusion states that "these advances may cause a rebound effect where efficiency gains cause an increase in demand for AI, resulting in an increase rather than a decrease in total resource use." "The AI enthusiasm of 2022 and 2023 may be part of such a rebound effect, and this enthusiasm points the way for AI server supply chains to make a greater contribution to global data center power consumption in the coming years."
This AI news is produced by the online assistant service "TIMEWELL."
Related Articles
- From full-time to part-time: the reality of working through two maternity leaves and how it changed my perspective on work | TIMEWELL
- Before taking parental leave: the 3 must-do points to ensure you can take leave even during busy season
- Embracing the role of a hands-on certified architect: finding my own path as the 5th generation of Fujita Construction