Currently, artificial intelligence (AI) is dominating different spaces, from technology to medicine. However, the more widely it is adopted, the larger the energy footprint it records. A new study has highlighted this and warned that AI’s energy footprint may exceed that of a small country in the near future.
In a commentary published October 10 in the journal Joule, author Alex de Vries warned that increased adoption of AI tools in the future could surge its energy footprints so much so that it could exceed the power demands of some countries, a Science Daily report said.
Since last year, especially after the launch of OpenAI’s ChatGPT, generative AI, which can produce text, images, or other data, has been used extensively to make everything – from coding to driving – less time-consuming. However, training these AI tools requires feeding the models a large amount of data, which is an energy-intensive process.
Elaborating on his point, de Vries points out that Hugging Face, an AI-developing company based in New York, has said that its multilingual text-generating AI tool consumed about 433 megawatt-hours (MWH) during training, which could power 40 average American homes for a year.
Even after training, when the AI tools generate data based on prompts, de Vries's analysis shows that it also uses a significant amount of computing power and hence, energy. For example, ChatGPT could cost 564 MWh of electricity a day to run, the Science Daily report explains. "The result of making these tools more efficient and accessible can be that we just allow more applications of it and more people to use it," de Vries adds in the commentary.
The production of AI servers is projected to grow rapidly in the near future. By 2027, global AI-related electricity consumption could increase by 85 to 134 terawatt-hours annually based on the projection of AI server production, the Science Daily report says. The amount is comparable to the annual electricity consumption of countries such as the Netherlands, Argentina, and Sweden.
Previously, a July 2023 study had highlighted AI models’ water footprint. The study published by the University of Colorado Riverside and the University of Texas Arlington revealed that “ChatGPT drinks 500ml of water" for every conversation.
As AI is the focus of many tech giants right now and is expanding to other fields, these studies warn about the long-term consequences of using it widely.