As the adoption of generative artificial intelligence (AI) technologies continues to rise, so does their environmental impact, warns Sasha Luccioni, a prominent researcher in the field. Recognized by Time magazine as one of the 100 most influential people in AI for 2024, Luccioni is on a mission to raise awareness about the substantial energy demands of generative AI models like ChatGPT and Midjourney.
At the ALL IN artificial intelligence conference in Montreal, Luccioni expressed her concerns about the energy consumption associated with these technologies. "Generative AI uses 30 times more energy than a traditional search engine," she noted, emphasizing the discrepancy between AI's energy demands and traditional data retrieval methods.
Generative AI models require vast computing resources to train on extensive datasets, involving powerful servers and significant energy usage. Unlike traditional search engines that retrieve existing information, AI models generate new content, which is inherently more energy-intensive.
According to the International Energy Agency, the combined AI and cryptocurrency sectors consumed nearly 460 terawatt hours of electricity in 2022, accounting for two percent of global electricity production. This substantial consumption highlights the pressing need for energy efficiency in AI development.
Luccioni, who co-developed the CodeCarbon tool in 2020 to quantify the carbon footprint of code execution, is now focusing on creating a certification system for AI algorithms. This system would allow developers and users to gauge the energy efficiency of AI models, similar to the energy efficiency ratings for household appliances provided by the US Environmental Protection Agency.
"We aim to measure and label the energy consumption of AI models," Luccioni explained. "This would help users make more informed decisions about which models to use, based on their environmental impact."
Despite her efforts, Luccioni faces challenges in obtaining data from commercial AI models from companies like Google and OpenAI, which have been reluctant to participate in transparency initiatives. Microsoft and Google have committed to achieving carbon neutrality by the end of the decade, but their greenhouse gas emissions surged in 2023 due to increased AI usage—up 48 percent for Google compared to 2019 and 29 percent for Microsoft compared to 2020.
Luccioni calls for greater transparency from tech companies and advocates for government intervention. "Governments are currently 'flying blindly' without knowing the specifics of data sets or algorithm training," she said. "Transparency is crucial for effective legislation and mitigating the climate crisis."
In her latest research, Luccioni demonstrated that generating a high-definition image with AI consumes as much energy as fully recharging a smartphone battery. As companies integrate AI into various aspects of daily life, Luccioni promotes the concept of "energy sobriety"—selecting and using AI tools judiciously to minimize their environmental impact.
"The goal is not to oppose AI but to choose the right tools and use them wisely," Luccioni emphasized. "We must balance technological advancement with environmental responsibility to ensure a sustainable future."