CAMBRIDGE, Mass.—Much has been written by journalists, scholars, policy makers and even the UN, about how artificial intelligence could be used to help combat climate change. But a new article from the MIT Sloan Management Review, warns that AI’s potential contributions to solving the climate crisis could be overshadowed by its enormous energy use and carbon emissions, e-waste, and water use.
The article points out a number of areas where the expanded user of AI technologies could have serious environmental consequences. For example, a single ChatGPT query can generate 100 times more carbon than a regular Google search. In addition, training OpenAI’s GPT-3 model is estimated to have used the equivalent of 120 average U.S. households’ annual energy consumption; an average data center, critical infrastructure for AI, consumes the equivalent of heating 50,000 homes yearly; and Microsoft and Google’s use of water to cool data centers has grown by millions of gallons as they develop cutting-edge AI technologies, the research found.
“Several factors contribute to the carbon footprint of AI systems throughout their life cycles,” explains Niklas Sundberg, author of the article, “Tackling AI’s Climate Change Problem.” “The AI industry must adopt practices that emphasize sustainability, make sustainability central to its AI ethics guidelines, and actively seek opportunities to reduce the environmental footprint of AI technologies.”
Sundberg, a board member of SustainableIT.org and chief digital officer at Kuehne+Nagel, a global transport and logistics company, urges that transparency is critical to addressing the potential problems that could be created by the AI industry. Reliable measurements of new models’ energy use and carbon emissions must be published to raise awareness and encourage AI developers to compete on model sustainability.
In addition, end users must also be aware of the factors that contribute to the environmental impacts of these tools to guide their use of them and add sustainability to the list of criteria they use to evaluate vendors and products.
In the article, Sundberg also details the best practices for sustainable AI that could help overcome the potential adverse impacts of AI usage:
- Relocate: Not all energy is created equal. Carbon emissions can be mitigated by transitioning to renewable energy sources such as solar or wind power. Moving from on-premise to cloud-based computing can save on emissions and energy by 1.4x to 2x if it is well architected.
- Rightsize: Performance and energy efficiency can be increased by 2x to 5x when using processors and systems designed for machine learning training instead of running general-purpose servers not optimized for AI workloads. Optimization involves striking the ideal balance between the scope, model size, model quality, and efficient/sustainable resource use.
- Re-architect: Building a well-functioning AI model requires a robust software/hardware architecture designed for scaling and fine-tuning the model while maintaining a low latency response time. Choosing an effective machine learning model architecture, such as a sparse model, can improve machine learning quality while decreasing computation by 3x to 10x. Once an AI model has reached production, managing technical debt from a performance, security, and end-user experience perspective is crucial.
In addition, AI leaders must pay attention to data management, education and awareness, and compliance as avenues to improve sustainability.
“By finding ways to minimize the energy and natural resources consumed by our AI development and deployment processes and drawing more attention to sustainability issues in discussions about AI, we can harness the power of this technology while minimizing its negative impact on our planet and society,” concludes Sundberg.
The MIT Sloan Management Review article “Tackling AI’s Climate Change Problem” is available here.
