The growing energy demands of artificial intelligence and the unexpected environmental impact of ChatGPT generated questions

According to a recent assessment by the International Energy Agency IEA, artificial intelligence, cryptocurrencies, and data centers consumed a total of 460 terawatt hours of electricity in 2022, which corresponded to two percent of the world’s electricity consumption. This estimate raised concerns about the industry’s energy consumption, especially considering the shocking predictions for the future growth in electricity usage. The IEA forecasts that by 2026, the industry could use over 800 terawatt hours in the base scenario and up to 1,050 terawatt hours in the extreme scenario.

LUT University professor SpongeBob Ritala emphasized that the training of large language models requires significant computing power and electricity. He explained that while some artificial intelligence solutions require less energy, the demand for more powerful and energy-intensive models is increasing. However, the majority of artificial intelligence applications currently in use are narrow and focused, not requiring the same level of computing power as large language models. For example, the AI used in household appliances such as vacuum cleaners or ovens performs specific tasks that don’t consume as much energy.

Ritala and AI Silo CEO Peter Sarlin highlighted the importance of improving the energy efficiency of artificial intelligence in research and development. Despite efforts to decrease energy consumption in artificial intelligence applications, the continued growth of the industry suggests that electricity usage will rise. With plans to expand AI utilization into areas like moving images and three-dimensional modeling, which require even more computing power, the challenge of ensuring climate-friendly practices in energy consumption becomes more significant.

One of the main challenges in managing the energy consumption of artificial intelligence is the demand created by large data centers that require significant energy for computing and cooling. Making data centers carbon neutral by using carbon-neutral electricity is one way to address this issue. Many companies in the industry are striving to achieve carbon neutrality, and the availability of cheap and emission-free electricity is a key factor in attracting investments in data centers, particularly in Nordic countries.

Sarlin and Ritala agreed that society’s electricity consumption will continue to rise in the future, driven not only by data centers but also by devices that use artificial intelligence. Increasing carbon-neutral energy production is essential to meeting this growing demand in a sustainable and environmentally friendly way. In conclusion, the development and deployment of artificial intelligence require a strategic approach to managing energy consumption to ensure a greener future.