Sustainability Concerns in Artificial Intelligence
As AI use grows rapidly, so do the environmental impacts. These concerns span energy consumption, water usage, emissions, and misinformation risks.
Energy Consumption
Data centers powering AI can consume as much electricity as small countries. The International Energy Agency (IEA) has estimated that the rise of AI could see data centers in Ireland account for nearly 35% of the country's energy use by 2026. It is estimated that a request through ChatGPT consumes ten times the electricity of a Google search.
The International Energy Agency (IEA) warns that AI’s global energy demand could triple by 2030 without intervention.
If powered by fossil fuels, AI development exacerbates climate change.
Water Usage
AI data servers use large volumes of water during construction and for cooling systems once the centers are operating.
This contributes to the depletion of freshwater resources, especially in drought-prone regions. This is a problem when a quarter of humanity already lacks access to clean water and sanitation.
The UN Environment Programme estimates that globally AI related infrastructure may soon consume six time more water than Denmark.
Disinformation and Climate Change Denial
The UN Environment Programme has reported that AI-generated content can be used to spread climate misinformation, so downplaying the threat.
Climate Action Against Disinformation (CAAD) highlight how generative AI may amplify false narratives that delay climate action. They call for regulation, transparency, safety and accountability to prevent AI from undermining global climate goals.
Resources
Building and maintaining AI infrastructure including data centers, chips and technology requires rare earth minerals and energy-intensive manufacturing.
The manufacture of a 2kg computer needs 800kg of raw materials according to the UN Environment Programme. Rare earth minerals are often mined in environmentally destructive ways.
AI servers produce electronic waste which can include hazardous substances such as mercury and lead.
Opportunities for Sustainable AI
There are hopes that AI can help tackle some of the world's environmental emergencies. The UN Environment Programme gives examples of AI being used to map destructive dredging of sand and to chart emissions of the greenhouse gas methane.
There is scope for efficiency improvements. Tech companies need to work on making AI algorithms more efficient to reduce energy demand and water use, recycling water and reusing components.
Green Data Centers: Transitioning to renewable-powered infrastructure.
Policy and Regulation: Governments and institutions must set sustainability standards for AI development. At the present time there is a lack of regulation according to Climate Action Against Disinformation (CAAD). The UN Environment Programme states that over 190 countries have adopted a series on non binding recommendations on the ethical use of AI which covers the environment.
Transparency. At the moment there is a lack of reliable information on the environmental impact of AI. There is a need to introduce standard procedures of measuring the environmental impact of AI which would include requiring companies to disclose the environmental impact of their AI systems.
Further Information and Reading
Warning AI industry could use as much energy as the Netherlands - BBC News
Climate Action Against Disinformation (CAAD) (2024) Artificial Intelligence Threats to Climate Change. Available at: https://foe.org/news/ai-threat-report
Energy and AI – Analysis - IEA
Explained: Generative AI’s environmental impact | MIT News | Massachusetts Institute of Technology
Environment and sustainability | The Alan Turing Institute
UN Environment Programme 21 September 2024 AI has an environmental problem. Here’s what the world can do about that.
AI Is Accelerating the Loss of Our Scarcest Natural Resource: Water