Syllabus: Conservation, environmental pollution and degradation, environmental impact assessment
Context and Significance
- Artificial Intelligence is expanding rapidly across healthcare, agriculture, governance, and industry sectors.
- Environmental consequences of AI development and deployment receive limited public and policy attention.
- An OECD working paper highlights carbon-intensive nature of AI compute systems.
Carbon and Energy Footprint of AI
- Global ICT sector contributes 1.8–2.8% of global greenhouse gas emissions, possibly higher.
- Some estimates place ICT emissions at 2.1–3.9% of global GHG output.
- Carbon footprint data of AI models is often incomplete or non-transparent.
- A 2025 Google report claimed one AI text prompt uses 0.24 watt-hours, attracting criticism.
- Critics argue such estimates ignore lifecycle and cumulative energy consumption.
Water Use and Lifecycle Impacts
- A 2024 issue note by United Nations Environment Programme warned of severe resource stress.
- AI servers may consume 4.2–6.6 billion cubic metres of water by 2027.
- Training one Large Language Model can emit around 300,000 kilograms of carbon dioxide.
- A 2019 study estimated 6,26,000 pounds of COâ‚‚ from training a single AI model.
- This equals the lifetime emissions of five passenger cars.
AI Usage and Comparative Energy Demand
- A 2024 UNEP study found ChatGPT queries consume ten times more energy than Google searches.
- Rising AI adoption increases aggregate electricity demand and climate risks.
Global Policy Responses
- In 2021, UNESCO adopted AI Ethics Recommendations, covering environmental harms.
- Around 190 countries adopted these non-binding guidelines.
- The United States and European Union proposed AI-specific environmental legislations.
Policy Imperatives for India
- India must recognise environmental costs of AI model development, not only benefits.
- Environmental Impact Assessment under EIA Notification, 2006 could be expanded to AI systems.
- Establish standardised metrics for emissions, energy, water, and resource consumption.
- Stakeholder engagement with tech firms, think tanks, and NGOs is essential.
Disclosure and Sustainability Measures
- AI environmental impacts can be integrated into ESG disclosure frameworks.
- The EU’s CSRD requires emissions disclosure from data centres and high-compute activities.
- Sustainable practices include pre-trained models, renewable-powered data centres, and transparent reporting.
