A recent study has reflected one of the most important concerns around artificial intelligence (AI): its carbon footprint. According to research published in Frontiers of Environmental Science & Engineeringthe growth and popularity of these systems is leading to alarming energy consumption. As more complex and robust models are developed, such as the case of OpenAI’s GPT-4, the energy required to operate them increases significantly, generating an environmental impact that continues to increase.
The training process of these models represents only a small part of the total energy consumption. However, the real challenge is in the continuous operation of AI tools, which, according to estimates, can consume up to 960 times more energy than a single training session.
This data is particularly worrying in a context where the use of AI is rapidly expanding in areas such as healthcare, transportation, education and entertainment, exponentially raising the global environmental impact of the technology.
Researchers and experts, such as Meng Zhang from Zhejiang University, warn that carbon emissions from the massive use of AI could become a huge problem, with annual costs that would exceed the 10 billion dollars if no action is taken. Zhang and his team point out the urgent need for governments and regulatory bodies to establish specific regulations to measure and reduce these emissions. Without standards that regulate the environmental impact of AI, the advancement of this technology could become a serious problem for sustainability.