Developers will need to find creative ways to mitigate energy consumption.
A new study published in the journal Joule highlights the potential impact of artificial intelligence (AI) technology on electricity consumption. The study warns that AI systems, particularly generative AI technology, heavily rely on powerful servers, which could lead to a significant surge in energy demand.
The study points to tech giant Google as an example, revealing that AI accounted for only 10-15% of the company’s total electricity consumption in 2021. However, as AI technology continues to expand, Google’s energy consumption could potentially reach the scale of a small country.
The authors of the study suggest that in a worst-case scenario, Google’s AI alone could consume as much electricity as a country like Ireland, which consumes approximately 29.3 TWh per year. This would represent a significant increase compared to Google’s historical AI-related energy consumption.
Experts argue that developers will need to find innovative ways to mitigate the energy consumption of AI technology. Christopher Alexander, the chief analytics officer of Pioneer Development Group, compares the demands of AI to Bitcoin mining and suggests that alternative energy sources, such as natural gas from oil drilling and biogas from landfills, could be utilized to lessen the environmental impact.
While concerns about the energy consumption of AI technology are valid, experts believe that improvements will likely be made to make energy usage more efficient. As with other growing technologies, initial inefficiencies are common, but as the industry evolves, chips and algorithms will be optimized, leading to lower energy usage levels.
It is important to temper both overly optimistic and overly pessimistic expectations for the future of AI technology and energy consumption. The study acknowledges that some extreme scenarios are unlikely to occur rapidly. However, it emphasizes the need for continued research and development to ensure that AI technology is sustainable and beneficial in the long run.
