If you're concerned about the environment, reconsider your use of AI technology. Generative artificial intelligence reportedly consumes 30 times more energy than standard search engines, as highlighted by researcher Sasha Luccioni, who is dedicated to raising awareness about the environmental consequences of this trendy technology. Named one of Time magazine's 100 most influential figures in AI in 2024, the Canadian computer scientist of Russian descent has been working for several years to measure the emissions produced by programs like ChatGPT and Midjourney. "I find it particularly disappointing that generative AI is being used for Internet searches, " laments Luccioni during an interview with AFP at the ALL IN artificial intelligence conference in Montreal. The language models underlying these AI applications require substantial computing power to process billions of data points, necessitating the use of high-performance servers. Moreover, there’s significant energy consumption involved in responding to individual user requests. Rather than simply retrieving information, "like a search engine would to find a country's capital, for example, " AI systems "generate new information, " making the process "much more energy-intensive, " she explains. According to the International Energy Agency, the AI and cryptocurrency sectors combined consumed nearly 460 terawatt-hours of electricity in 2022—accounting for two percent of total global electricity production. - Energy Efficiency - A leading authority on AI's environmental effects, Luccioni co-developed a tool in 2020 that enables developers to measure the carbon footprint of their code. "CodeCarbon" has been downloaded over a million times since. Currently heading the climate strategy at Hugging Face, a platform for sharing open-access AI models, she is also focused on establishing a certification system for algorithms. Similar to the program from the US Environmental Protection Agency, which rates electronic devices based on energy consumption, this system aims to inform users and developers about an AI product's energy usage, encouraging them to "make better choices. " "We may not account for water usage or rare materials, " she concedes, "but we can measure energy efficiency for specific tasks and provide grades, such as A+ for one model and D for another. " - Transparency - To develop her tool, Luccioni is testing it on publicly accessible generative AI models.

However, she seeks to evaluate commercial models from Google or OpenAI, both of which have been hesitant to cooperate. Though Microsoft and Google have pledged to achieve carbon neutrality by the end of the decade, both tech giants witnessed significant increases in greenhouse gas emissions in 2023 due to AI: Google’s emissions rose by 48 percent compared to 2019, while Microsoft’s increased by 29 percent since 2020. "We are accelerating the climate crisis, " warns Luccioni, urging greater transparency from technology corporations. She believes that governments could provide a solution, as they are presently "flying blindly, " unaware of the contents of data sets or the training of algorithms. "Once transparency is established, we can begin to legislate. " - 'Energy Sobriety' - Luccioni also emphasizes the need to "educate people on what generative AI can and cannot achieve, and at what expense. " Her recent research demonstrated that generating a high-definition image with AI uses as much energy as fully charging a smartphone battery. As more companies strive to incorporate AI into everyday life—through conversational bots, connected devices, and online searches—Luccioni advocates for "energy sobriety. " She clarifies that the goal isn’t to reject AI but to select the appropriate tools and employ them wisely.