Think Before Using AI – Environmental Impact
Image created by Khyati Trehan as part of the Visualising AI project launched by Google DeepMind.
Artificial Intelligence (AI) has huge potential to drive life-changing developments, particularly in addressing the world's environmental challenges. Yet, behind this excitement lies a growing concern about the environmental impact of AI technologies. These expansive networks are hosted in electricity-hungry, resource-intensive data centres that pose substantial concerns for our planet and their long-term environmental impact is unknown.
In this blog, we'll explore the environmental implications of AI to encourage consideration before implementing and using AI systems.
How Can AI Benefit the Environment?
Before diving into the negatives, it’s important to recognise the incredible potential AI offers in environmental monitoring. If the negative impacts can be mitigated AI could play a key role in tackling one of the world’s toughest challenges – climate change.
AI’s ability to detect and analyse patterns is making it an invaluable tool in environmental monitoring. Some of the many cases it is currently being used in include:
Iceberg Monitoring: AI is being trained to track iceberg movements and melting rate, tasks that are typically challenging and time-consuming. Using AI would allow for faster identification of melting patters allowing for improved actions to be taken as the amount of meltwater in the ocean is better understood.
Deforestation Mapping: By combining AI with satellite imagery, ecology experts can map deforestation and carbon stored in forests. Scottish company Space Intelligence is currently working on this.
Weather Prediction: AI can predict weather patterns, helping authorities to better prepare for extreme weather conditions.
Sustainable Agriculture: AI is improving efficiency in agriculture through soil quality and crop analysis.
Waste Management and Recycling: AI can improve waste management and recycling from offering improved waste collection routes to enhanced material sorting at recycling facilities.
These are just some of the many examples of how AI could contribute to environmental monitoring.
The Negative Environmental Impact of AI
Despite the potential benefits, AI also poses significant environment risks. As AI demand grows, so too does its environmental footprint. Balancing AI’s advantage with its costs is critical to protecting our planet.
Let’s explore some of the key concerns:
Scarce Recourse Use: Some AI systems require rare earth elements, which are already scarce. Not only does it put pressure on limited supply but the mining for these elements can also be extremely destructive to the environment.
Electronic Waste: AI systems are often developed and maintained in data centres which generate huge amounts of electronic waste. This waste contains hazardous substance which can pollute the air if not disposed of properly.
Water Consumption: A lot of water is consumed throughout a data centre’s life, from construction to everyday operation. This heavy usage is concerning given that access to clean water is already limited in some parts of the world.
Energy Consumption / Greenhouse Gas Emissions: AI technology requires huge amounts of energy and, if not sourced from renewables, they contribute to greenhouse gas emissions. Researchers from the University of Massachusetts found that training an AI model can emit around 300,000kg of carbon dioxide, equal to roughly 300 round-trip flights between Ney York and San Francisco.
Data Storage: AI require vast amounts of storage which requires substantial energy and water, placing additional strains on resources.
Impact on Vulnerable Populations: These environmental consequences often hit the most vulnerable communities the hardest, particularly in areas where these resources are already scarce.
Managing AI’s environmental impact will be challenging. Raising awareness is essential to encourage governments to develop of strategies and policies that will minimise AI’s environmental damage.
How Can We Counteract These Negatives?
To get the best of both worlds we must find a way in which we can harness AI’s benefits while minimising its harm to the planet. Some potential solutions include:
Data Centre Improvement: Data centres should be designed in an eco-friendly manner, prioritising energy efficiency and renewable energy sources. Scotland has been established as an excellent location for green data centres due to its climate. The Scottish Government currently has plans to position Scotland as a leading in green data centres. Find out more here.
Heat/Carbon Offsetting: Data centres can participate in offsetting programmes which repurpose the energy they generate. There have already been examples of data centres being used to heat local student accommodation and pools.
Renewable Energy Sourcing: Switching to renewable energy sources will hugely reduce AI’s carbon footprint.
By focusing on environmentally friendly AI system development, and considering when AI is truly needed, we can reduce its impact and harness its true potential.