Influential AI researcher Sasha Luccioni has raised significant concerns about the vast energy consumption associated with the growing use of artificial intelligence, particularly generative models like ChatGPT and Midjourney.
Recently recognized by Time magazine as one of the top 100 influential figures in AI, Luccioni is raising awareness of these technologies’ environmental impact.
At the ALL IN Artificial Intelligence conference held in Montreal, Luccioni explained that generative AI models are far more energy-intensive than traditional search engines.
According to her research, models like ChatGPT use up to 30 times more energy than search engines because they generate new content in response to user prompts instead of merely retrieving information. This difference significantly increases their power demands.
A key factor behind this high energy consumption is the extensive computational resources required to train AI models on vast datasets. Once trained, these models continue to use substantial electricity as they operate, responding to user inputs.
Luccioni emphasized that these powerful AI systems rely on energy-intensive servers to process and manage large amounts of data, a primary driver of their environmental impact.
The International Energy Agency has also noted AI’s increasing strain on global energy resources. According to their estimates, AI and cryptocurrency consumed approximately 460 terawatt hours of electricity in 2022, amounting to 2% of the world’s total electricity production. This figure is expected to rise as AI technologies become more widely adopted.
In response to these issues, Luccioni has been actively working to promote tools and strategies to mitigate AI’s environmental footprint. In 2020, she helped develop a tool called CodeCarbon, which allows developers to measure the carbon footprint of their code. Over one million people have used this tool, according to a report from AFP.
Luccioni’s latest project involves creating a system to rate AI models based on their energy efficiency. This would work similarly to energy-efficiency labels on household appliances like refrigerators.
Despite ongoing efforts by major tech companies like Google and Microsoft to address their environmental impact, AI continues contributing to rising greenhouse gas emissions. Both companies have made public commitments to become carbon-neutral by the end of the decade, but their emissions have increased significantly in recent years.
Google’s emissions have surged by 48% since 2019, while Microsoft’s have risen by 29% since 2020, according to the AFP report.