AI and the Environment: The Hopes, Risks, and How to Move Forward

AI is already being used to monitor methane emissions, track harmful sand dredging, and even predict future environmental outcomes by analysing data patterns. This technology could support governments and organizations in making eco-friendly decisions, giving hope for solutions to climate change, biodiversity loss, and pollution.

Is the environment ready for the impact of AI? The Environmental Cost of AI

1. Resource-Intensive Data Centres

AI needs vast data centres, which require enormous amounts of water, electricity, and minerals. The construction and operation of these facilities are resource-heavy; for instance, building a 2 kg computer uses 800 kg of raw materials.

2. Electronic Waste and Hazardous Chemicals

Data centres produce a lot of electronic waste, often containing toxic substances like mercury and lead, which harm the environment when improperly disposed of.

3. High Energy Consumption

AI processing is energy-intensive, often powered by fossil fuels, which increases greenhouse gas emissions. An AI request, such as one made to ChatGPT, uses 10 times the electricity of a standard Google search.

The Unpredictable Impact of AI

AI’s environmental impact can also be hard to predict, making it something of a wildcard. For example, AI-powered self-driving cars could encourage more car use, increasing emissions. AI could also be misused, such as in spreading misinformation about climate change, misleading public opinion on environmental risks.

What’s Being Done About It?

Some regulations are beginning to address AI’s environmental effects. Over 190 countries have adopted ethical guidelines for AI, covering environmental aspects, and the EU and U.S. have introduced legislation to reduce AI’s environmental footprint. However, more targeted policies are needed.

UNEP’s Recommendations for Sustainable AI

To mitigate AI’s impact, the United Nations Environment Programme (UNEP) suggests five key actions:

1. Standardized Measurements
2. Develop standard methods to measure AI’s environmental impact for more consistent data.
3. Transparency in Environmental Impact
4. Require companies to disclose AI’s direct environmental consequences to promote accountability.
5. Efficient AI Algorithms
6. Tech companies should make AI algorithms more energy-efficient and recycle water and components wherever possible.
7. Greener Data Centres
8. Encourage the use of renewable energy in data centres and carbon offset programs to lessen AI’s carbon footprint.
9. Integrating AI with Environmental Policy
10. Align AI policies with broader environmental regulations to create a sustainable framework for AI deployment.