Generative AI uses 30 times more energy than a traditional search engine
Used artificial intelligence Generative AI has 30 times the power of a traditional search engine, it warns. Researcher Sasha Lucioni which seeks to raise awareness of the environmental impact of this new technology.
For years, this Canadian researcher, who was ranked by Time magazine as one of the 100 most influential people in the world in 2024, has been trying to determine the amount of emissions produced by programs such as ChatGPT and Midjourney.
“I find the use of generative AI to search the internet disappointing,” Lucioni told AFP in an interview on the sidelines of the AllIn AI conference in Montreal.
The language models that AI programs rely on require massive computing power to train on billions of data points, which requires powerful servers, in addition to the energy consumed to respond to user requests.
She explained that instead of simply extracting information, “like a search engine would do to find the capital of a country, for example,” AI programs “generate new information,” making the process “more energy-intensive.”
The International Energy Agency reported that the AI and cryptocurrency sectors consumed about 460 terawatt hours of electricity in 2022, or 2% of total global production.
Energy efficiency
In 2020, Lucioni, a leading researcher on the impact of AI on Climatein creating a tool called CodeCarbon for developers to determine the carbon footprint left by running a piece of code.
Lucioni, who heads climate strategy at Hugging Face, a startup that shares open-access AI models, is currently working on creating a certification system for algorithms.
Similar to the US Environmental Protection Agency’s Energy Star program, which awards points to electronic devices based on how much energy they consume, this system will enable users and developers to know how much energy an AI product consumes in order to encourage users and developers to “make better decisions.”
transparency
To develop her tool, Lucioni is testing it on publicly accessible generative AI models, but she also wants to test it on commercial models from Google and OpenAI, the company that developed ChatGPT, which has so far been reluctant to agree.
Despite Microsoft’s commitment And “Google” Achieving carbon neutrality by the end of the decade, the two American companies saw their greenhouse gas emissions rise significantly in 2023 due to AI: a 48% increase for Google compared to 2019 and a 29% increase for Microsoft compared to 2020.
“We are accelerating the climate crisis,” Lucioni said, calling for more transparency from tech companies.
The solution, she added, could come from governments that are currently “driving blind,” perhaps without knowing what is “in the data sets or how the algorithms are trained.”
“Energy Sobriety”
The researcher noted that it is also necessary to “explain to people what generative AI can and cannot do, and at what cost.”
In her latest study, Lucioni showed that producing a high-resolution image using AI consumes as much energy as it takes to recharge a cell phone battery to its maximum capacity.
At a time when more companies want to integrate technology more deeply into everyday life, through chatbots, connected devices or online searches, Lucioni called for “energy sobriety.”
She stressed that “the idea here is not to oppose artificial intelligence, but to choose the right tools and use them wisely.”
