AI's energy appetite: The rising electricity costs of AI use
Artificial intelligence finds applications in many fields, from facilitating daily tasks and supporting education to entertainment. However, using AI comes with specific costs, especially regarding electricity consumption. What are these costs?
24 May 2024 16:02
The popularity of artificial intelligence is growing. Since the ChatGPT application became widely available, more and more people have been using AI tools for both professional and recreational purposes. However, generating a funny image, creating a song, or solving a complicated equation requires significant electricity.
As much energy as Japan
According to swiatoze.pl, based on data from the International Energy Agency (IEA), in 2022, cryptocurrencies and artificial intelligence accounted for about 2 percent of global electricity consumption. Forecasts indicate that by 2026, this share could double.
This means that in two years, artificial intelligence and cryptocurrencies will consume as much electricity as Japan currently needs. The reason for this increase is clear.
Why does AI need so much energy?
Generative artificial intelligence, which creates texts, images, or sounds based on text commands, must perform complex calculations. This process results in the expected content, but first, AI needs to be "fed" with the necessary information.
To perform these calculations, powerful servers are required, which in turn need significant amounts of electricity to operate. AI tool developers strive to increase their efficiency and shorten response times, which further translates into higher energy consumption—the demand for energy increases by about 26 to 36 percent annually.
Artificial intelligence might require even more significant amounts of electricity in the future. Journalists from swiatoze.pl emphasize that AI tools operate in two stages. The first is "feeding," or training the program based on data. This stage consumes about 20 percent of the total required energy. The second stage is usage, where people give commands and expect AI to produce results, consuming 80 percent of energy.
Given artificial intelligence's growing popularity and capabilities, understanding its impact on electricity consumption becomes crucial for the future of energy resource management strategies.
