AI/automation ethics glossary
Environment
AI/automation ethics glossary
Environment
Environment refers to the development, deployment, or operation of an AI/automated system in such a way that it damages the environment through excessive energy consumption, resource depletion, pollution, or other actions.
AI systems can be resource-intensive at every stage: mining minerals for hardware, manufacturing chips and servers, powering data centers, cooling equipment with water, training models, serving user queries, and eventually disposing of outdated devices.
Generative AI is especially demanding because models are often trained and updated repeatedly, which increases electricity and water use.
The ethical issue is not only whether AI works, but whether its environmental footprint is justified by the value it creates.
AI's environmental footprint intersects with the most urgent global challenge of our era: climate change.
On the one hand, there are high hopes that AI can help tackle some of the world's biggest environmental emergencies; it is already being used to map the destructive dredging of sand and chart emissions of methane.
On the other hand, as AI becomes central to the global economy, its environmental cost threatens to undermine international climate goals. If the growth of AI outpaces the transition to renewable energy, the technology could accelerate ecological collapse rather than solve it.
Ethical stewardship requires that the benefits of AI and automation are not erased by the physical harm done to the planet.
The negative impacts of AI and automation systems on the environment include:
Air pollution. The contamination of the air by harmful substances that change its natural composition and make it unsafe or unpleasant for living beings and the environment.
Ground pollution. The contamination of soil and subsurface by harmful substances that damage ecosystems and human health.
Water pollution. The contamination of natural water bodies by substances or forms of energy that make the water unsafe for living organisms or human use.
Noise pollution. An unwanted or unwarranted sound that can disturb human and animal health and wellbeing.
Excessive water consumption. Leading to water restrictions and shortages in local communities and amongst local farmers.
Excessive energy shortages. Excessive energy use resulting in energy bottlenecks and shortages for communities, organisations and businesses.
Climate change acceleration. High CO₂ emissions from non-renewable energy grids used by data centers contribute to extreme weather patterns.
Ecological/biodiversity loss. Deforestation, habitat destruction and the fragmentation and loss of biodiversity due to the over-expansion of technology infrastructure, or inadequate alignment of technology with sustainable practices.
Resource depletion. Intensive mining for lithium and cobalt destroys local ecosystems and depletes non-renewable resources.
Environmental inequality. Environmental harms often disproportionately affect developing nations (where mining and e-waste dumping occur), while the technological benefits accrue to wealthier nations.
The sources of environmental damage caused directly or indirectly by AI and automation systems are many and varied:
Rapid AI scaling. The race to build ever-larger models prioritises performance over resource efficiency
Fossil fuel dependency. Many data centres remain powered by coal or gas grids, especially in high-growth regions.
Insufficient regulation. Few jurisdictions mandate environmental impact assessments for AI infrastructure.
Opaque corporate reporting. Companies aggregate AI and non-AI workloads, obscuring the true environmental cost.
Water-energy tradeoffs. Optimising for energy efficiency (e.g., evaporative cooling) often increases water consumption.
Geographic concentration. Clustering of data centers in already water- or energy-stressed regions amplifies local harms, including excessive water and energy consumption, and water and ground pollution.
Hardware churn. The rapid obsolescence of AI chips generates large volumes of e-waste - an early signal that internal accountability mechanisms were insufficient to surface environmental concerns within major AI labs.
Innovation versus conservation. Is the potential for AI to solve climate change (e.g., optimising energy grids) worth the massive carbon footprint required to develop it?
Access versus sustainability. Should researchers in resource-poor regions be limited in their AI development to meet global green standards, or does this stifle global equity?
Speed versus efficiency. Should companies prioritise being first to market, or delay releases to optimize algorithms for lower energy consumption?
Fairness
Human rights & civil liberties
You are welcome to use, copy, adapt, and redistribute this definition under a CC BY-SA 4.0 licence.
Let us know if you have any comments or suggestions about how to improve this definition, or would like to suggest and/or contribute additional terms to define.
Author: Charlie Pownall 🔗
Published: May 10, 2025
Last updated: May 10, 2025