Sustainability in the Age of Artificial intelligence (AI)
By Aubrey Kemper, Environmental Manager (Chicago, Illinois)
Aubrey Kemper holds a Bachelor's degree in Coastal Environmental Science from Flagler College in St. Augustine, Florida, and a Master's degree in Natural Resources from University of Nebraska (Lincoln). Her greatest passion is environmental education for all ages, and much of her academic research involved creating educational resources for "citizen scientists". Before coming to Vanguard Environmental, Inc., she served as a nature educator at Lyman Woods Nature Center in Downers Grove, Illinois. Now, that experience is channeled towards educating her ten nieces and nephews to become great stewards of the environment.
In October 1950, renowned computer scientist and mathematician Alan Turing published “Computing Machinery and Intelligence”, in which he introduced what is now known as the “Turing test” for determining the ability of a machine to answer questions in a way which resembles a human language model. In the 75 years since the publication of this seminal work, numerous artificial intelligence (AI) models have passed variations of the Turing test. Now that the technology exists and is being widely implemented, how can we use it responsibly?
One of the biggest concerns regarding widespread use of AI is that the systems required to power and maintain the programs require massive amounts of energy. For example, a medium-sized generative AI model employing a technique known as ‘neural architecture search’ has an energy consumption equivalent to 626,000 tons of CO2 emissions. To put that into perspective, that is the same amount of CO2 emissions created by driving five average American cars over their lifetime. Larger-scale models require data centers in order to operate. Data centers not only require even more energy, but they are producers of hazardous substances such as mercury and lead. Likewise, data centers require large amounts of water to cool the electrical components.
On the other hand, AI can be used to detect patterns in environmental data. For example, AI programs have been used to graph trends in greenhouse gas emissions for the oil and gas industry. This allows governments to predict the environmental impact of certain industries and develop regulations to mitigate potential harmful effects. Similar models are used to track wildlife migration patterns, vulnerable and endangered species populations, deforestation rates, and climate data. These programs are able to detect small anomalies in the collated data and determine indicators of larger environmental concerns.
One way for environmentally conscious AI developers to minimize their consumption is to focus on energy efficiency and resource conservation. Such companies utilize less energy consumptive machinery, optimize the energy that is consumed, and use alternatives to fossil fuels wherever possible. There is no doubt that AI is in the midst of a renaissance period. However, it is important to consider the long-term environmental impact of large-scale operations. If you want to determine your facility’s potential environmental “weak spots”, ask your Regulatory Specialist about Environmental Facility Compliance Assessments (FCA-E) today!
References
- A. M. Turing (1950) Computing Machinery and Intelligence. Mind 49: 433-460.
- IFC (April 30, 2007). Environmental, Health, and Safety General Guidelines.
- Bashir, N., Donti, P., Cuff, J., Sroka, S., Ilic, M., Sze, V., Delimitrou, C., & Olivetti, E. (2024). The Climate and Sustainability Implications of Generative AI. An MIT Exploration of Generative AI. https://doi.org/10.21428/e4baedd9.9070dfe7.
- Terra, J. (2024, April 18). What is Sustainable AI? Definition, Significance, and Examples. Caltech -. https://pg-p.ctme.caltech.edu/blog/ai-ml/what-is-sustainable-ai-significance-examples