MONTREAL: If you care about the environment, think twice about using AI. Generative artificial intelligence uses 30 times more energy than a traditional search engine, warns researcher Sasha Luccioni, whose aim is to raise awareness of the new technology’s impact on the environment. The Canadian computer scientist of Russian origin, whom the American magazine Time ranked among the 100 most influential people in the world of artificial intelligence in 2024, has been trying to quantify the emissions of programs such as ChatGPT or Midjourney for several years. . “I find it particularly disappointing that generative artificial intelligence is being used to search the Internet,” complained the researcher, speaking to AFP on the sidelines of the ALL IN artificial intelligence conference in Montreal. The language models on which the programs are based require enormous computing power to train on billions of data points, which requires powerful servers. Then there is the energy used to respond to the demands of each individual user. Instead of simply retrieving information “like a search engine would do to find a country’s capital,” AI programs “generate new information,” making the whole thing “much more energy intensive,” he explains. According to the International Energy Agency, the combined AI and cryptocurrency sectors consumed nearly 460 terawatt-hours of electricity in 2022 – two percent of total global production. In 2020, Luccioni, a leading researcher on the impact of artificial intelligence on the climate, participated in the creation of a tool for developers to quantify the carbon footprint of running certain code. “CodeCarbon” has since been downloaded over a million times. The head of climate strategy at startup Hugging Face, an open-access platform for sharing artificial intelligence models, is now working to create a certification system for algorithms. Similar to the US Environmental Protection Agency’s program that awards scores based on the energy consumption of electronic devices and appliances, AI would allow knowing a product’s energy consumption and encourage users and developers to “make better decisions.” “We don’t consider water or precious materials,” he acknowledges, “but at least we know that for a particular task we can measure energy efficiency and say this model has an A+ and this model has a D,” he says. . To develop her tool, Luccioni is experimenting with it on publicly available or open source generative AI models, but she would also like to do so on commercial models from Google or OpenAI creator ChatGPT, which have been reluctant to agree. Although Microsoft and Google have pledged to become carbon neutral by the end of the decade, the US tech giants saw an increase in greenhouse gas emissions due to artificial intelligence in 2023: 48 percent for Google compared to 2019 and 29 percent for Microsoft compared to 2020. “We are accelerating the climate crisis,” Luccioni says, calling for more transparency from tech companies. The solution, she said, could come from governments, which at the moment are “flying blind” without knowing “what’s in the data sets or how the algorithms are trained.” “Once we are transparent, we can start legislation.” According to Luccioni, it is also necessary to “explain to people what generative artificial intelligence can and cannot do and at what cost.” In her latest study, the researcher proved that creating a high-resolution image using artificial intelligence consumes as much energy as fully charging your cell phone battery. At a time when more and more companies want to further integrate this technology into our lives – through chatbots and connected devices, or in online search – Luccioni advocates “energy sobriety”. The idea, he stresses, is not to stand up to AI, but rather to choose the right tools — and use them judiciously.
