The rapid growth in artificial intelligence has opened new opportunities while also raising novel concerns about its implications. One of such is the supposed positive and negative impacts of artificial intelligence on the environment. An emerging area of research has focused on examining these impacts with the intent of maximizing the positive environmental impact of AI and minimizing or eliminating its negative environmental impact.
Environmental and Technological Paradox: The Dual Environmental Impact of Artificial Intelligence
Advances in technology have always created dilemmas. The same tech or innovation that has been viewed to provide beneficial effects has also been blamed for unfavorable tradeoffs. The field of artificial intelligence is not a stranger to these so-called technological paradoxes. It has created new opportunities and has been seen to open new possibilities with its further development but it has also been scrutinized for its purported offshoots. The dual environmental impacts of artificial intelligence best represent a technological paradox.
Positive Environmental Impacts of Artificial Intelligence
1. Improving Energy Efficiency
Several studies and even real-world applications have used AI to solve pressing environmental issues and problems. Specific algorithms and models have been used in various energy management solutions aimed at optimizing energy consumption, reducing usage during peak hours, determining signal problems, detecting failures even before they occur, and improving the integration and utilization of renewable and intermittent energy sources.
Researchers J. Liu et al. studied how specific AI applications help manufacturing companies in China improve their energy consumption. Results showed that using industrial robots can result in significant energy efficiency while also improving technological efficiency. Take note that tech companies such as Amazon and Google use machine learning and deep learning algorithms to optimize and reduce their energy consumption in their data centers.
Power companies are also using AI to manage power grids. Specific algorithms can help in determining peak energy consumption and in predicting when renewable energy will be available. South Korean conglomerate SK Group has made significant investments in energy storage systems that can automatically pinpoint when to offload stored energy from renewable sources and into the power grid using artificial intelligence.
2. Enhancing Environmental Monitoring
Assessing the environment is an important aspect of environmental management and conservation programs. Environmental monitoring is specifically a systematic process that involves collecting, analyzing, and interpreting data and information about the environment. Its purpose is to understand and evaluate the state of the environment, identify changes or trends, and assess the impact of human activities or natural processes.
The World Environment Situation Room or WESR of the United Nations Environment Programme is leveraging the capabilities of artificial intelligence in analyzing complex and multifaceted datasets. It specifically curates, aggregates, and visualizes the best available earth observation and sensor data to inform near real-time analysis and future predictions on multiple factors such as glacier mass, sea level rise, and greenhouse gas emissions.
Artificial intelligence also advances the field of climate informatics. This novel discipline that emerged in 2011 from the combined principles and practices in data science and climate science covers a multitude of topics ranging from improving the prediction of extreme events such as hurricanes to predicting weather at a hyper-local level and making bold assumptions about the social and economic impacts of weather and climate.
3. Supporting Environmental Management
Researchers D. Rolnick et al. suggested 13 areas in the realms of environmental science and conservation programs in which machine learning can be deployed. These include the design of more energy-efficient buildings, the creation of new low-carbon materials, forest management and better monitoring of deforestation, implementation of precision agriculture and monitoring of lands for livestock use, and design of planetary control systems, among others.
Combining satellite imagery with artificial intelligence can help in detecting changes in land use, vegetation, forest cover, and fallout from natural disasters. Invasive species can also be monitored and tracked. Anti-poaching units and conservation enforces can maximize the advantages of using the predicting powers of artificial intelligence. These are all important in conservation efforts and maintaining or improving biodiversity.
The management of waste and other pollution can be improved using machine learning algorithms. A specific use case involves the deployment of a system designed for tracking levels of pollution in the air and prompting concerned agencies to take relevant actions. Algorithms can also be used in improving different waste management processes such as the optimization of collection routes and determining gaps in relevant law enforcement.
Negative Environmental Impacts of Artificial Intelligence
1. Increasing Total Energy Consumption
Remember that artificial intelligence can help in reducing energy consumption or improving energy efficiency through better energy management. However, when it comes to the development and deployment of different architectures, algorithms, and models for different AI applications, artificial intelligence is an energy-intensive undertaking. Recent advances in the field have uncovered its energy requirement and energy utilization level.
The European Commission of the European Union explains that the data centers that store and process algorithms and models use a lot of energy. The collective energy consumption of activities and pursuits related to the field is even compared with the fuel consumption of the aviation industry. Take note that energy consumption comes from powering relevant hardware components such as processors and running cooling systems.
Google is an interesting example. Researchers found out that AI made up 10 to 15 percent of its total electricity consumption in 2021. This was equivalent to 18.3 terawatt hours. Another notable example is the AI models from OpenAI. The training of its GPT-3 large language model consumed about 284 megawatt-hours of energy. This is roughly equivalent to the annual energy consumption of 25 average households in the United States.
2. Contributing to Greenhouse Gas Emissions
A paper penned by Karen Hao and published in the MIT Technology Review mentioned that cloud computing has now a larger carbon footprint than the entire airline industry while a single data center might consume an amount of electricity equivalent to 50000 homes. Training a single AI model can emit more than 626000 pounds of carbon dioxide equivalent. This is about five times the lifetime emissions of an average-sized automotive vehicle.
Stanford University also released a report that evaluated the carbon footprint of three large language models. These were Gopher from Google DeepMind, OPT from Meta Platforms, BLOOM from Big Science, and GPT-3 from OpenAI. GPT-3 had the biggest footprint. The training of this model released 502 metric tons of carbon. Note that the GPT family of transformer-based language models powers the popular ChatGPT chatbot.
Key movers in the field of artificial intelligence have adopted a mindset centered on size and scale. This has been demonstrated through the release of deep learning models, specific large language models, and general multipurpose foundation models. These models are trained with huge datasets or big data and involve billions of parameters. The bigger the model the greater its energy requirement both for training and end-user utilization.
3. Contributing to Electronic Waste Generation
One of the challenges in advancing the field of artificial intelligence and the development and deployment of experimental and practical AI systems is the need to build and maintain technological infrastructure. Organizations involved in the development of systems need to have access to processing units such as discrete graphics processors and tensor processors, large storage mediums, networking infrastructure, and specific security solutions.
Developments in the field of AI and the expansion of AI applications are increasing the demand for semiconductors and specialized processors called artificial intelligence accelerators with tailor-fitted instruction set architectures. Several chipmakers including Apple, Google, Qualcomm, Nvidia, and Intel are designing chips or chip components tailor-fitted to handle the processing of machine learning algorithms and running specific AI models.
However, because of the advancements in artificial intelligence, the field is contributing to the generation of electronic waste. Electronic devices and components now have shorter lifespans because the rapid pace of technological development renders them obsolete. Experimental and even end-use hardware for data centers and specific AI applications such as robotics are also increasing the high turnover of hardware components.
FURTHER READINGS AND REFERENCES
- Ekin, A. 2019. “AI Can Help Us Fight Climate Change. But It Has an Energy Problem, Too.” Horizon: The Eu Research and Innovation Magazine. European Commission, European Union. Available online
- Hao, K. 2019. “Training a Single AI Model Can Emit as Much Carbon as Five Cars in Their Lifetimes.” MIT Technology Review. Available online
- Liu, J., Qian, Y., Yang, Y., and Yang, Z. 2022. “Can Artificial Intelligence Improve the Energy Efficiency of Manufacturing Companies? Evidence from China.” International Journal of Environmental Research and Public Health. 19(4): 2091. DOI: 3390/ijerph19042091
- Rolnick, D., Donti, P. L., Kaack, L. H., Kochanski, K., Lacoste, A., Sankaran, K., Ross, A. S., Milojevic-Dupont, N., Jaques, N., Waldman-Brown, A., Luccioni, A., Maharaj, T., Sherwin, E. D., Mukkavilli, S. K., Kording, K. P., Gomes, C., Ng, A. Y., Hassabis, D., Platt, J. C., … Bengio, Y. 2019. “Tackling Climate Change with Machine Learning.” arXiv. DOI: 48550/ARXIV.1906.05433
- Saul, J. and Bass, D. 2023. “Artificial Intelligence Is Booming—So Is Its Carbon Footprint.” Bloomberg. Available online
- Stanford University Human-Centered Artificial Intelligence. 2023. Measuring Trends in Artificial Intelligence: The AI Index Report. Stanford University. Available online
- SK Group. 2022. “Artificial Intelligence Can Bring Real Solutions to the Power Grid Crisis.” Forbes. Available online
- United Nations Environment Programme. 2022. “How Artificial Intelligence is Helping Tackle Environmental Challenges.” Climate Action. United Nations Environment Programme. Available online
- Wang, E.-Z., Lee, C.-C., and Li, Y. 2022. “Assessing the Impact of Industrial Robots on Manufacturing Energy Intensity in 38 Countries.” Energy Economics. 105: 105748. DOI: 1016/j.eneco.2021.105748