GS Paper: GS3

Why in the News?

Artificial Intelligence is expanding rapidly across sectors. However, its environmental costs remain largely ignored in policy discussions. The global ICT sector contributes 1.8-2.8% of global greenhouse gas emissions, with estimates rising to 2.1-3.9%. For the first time, clear data is available on the energy, water, and carbon footprint of AI systems, including Large Language Models (LLMs).

A clear gap exists between perceived digital efficiency and actual environmental impact. A single ChatGPT query consumes 10 times more energy than a Google search. Training one LLM can emit up to 3,00,000 kg of carbon dioxide. Despite these costs, India has no formal system to measure or disclose AI’s environmental impact. This contrasts with the EU and the US, highlighting a major governance gap.

What is the scale of AI’s environmental footprint?

1. Global ICT emissions: Accounts for 1.8-2.8% of global GHG emissions, with upper estimates reaching 3.9%.
2. Carbon-intensive training: Training a single LLM can emit ~3,00,000 kg of carbon dioxide.
3. Comparative impact: Emissions from one deep learning model equal emissions from five cars over their lifetime.
4. Data gap: Carbon footprint data of AI models and users remains fragmented and inconsistent.

How does AI affect energy consumption patterns?

1. High energy intensity: Each ChatGPT query consumes 10× more energy than a Google search.
2. Hidden electricity demand: AI workloads rely on energy-intensive data centres and specialised hardware.
3. Misleading averages: Claims such as 0.24 watt-hours per AI query underestimate system-wide consumption.

Why is water consumption emerging as a major concern?

1. UNEP projection: AI data centres may consume 4.2-6.6 billion cubic metres of water by 2027.
2. Cooling requirements: Water is extensively used to cool AI servers.
3. Water security risks: High freshwater withdrawal threatens water-stressed regions.

What global governance responses are emerging?

1. UNESCO framework (2021): Recognises negative environmental impacts of AI; adopted by ~190 countries.
2. European Union leadership:
   1. AI Act, 2024: Introduces environmental accountability in AI governance.
   2. Harmonised AI rules: Address sustainability alongside ethics and safety.
3. United States approach: Sector-specific regulations addressing AI’s environmental externalities.

Why does India need a regulatory shift?

1. Unaccounted externalities: Environmental costs of AI development remain outside policy evaluation.
2. Regulatory vacuum: No mandatory assessment of AI’s environmental impact.
3. Climate obligations: AI expansion risks undermining India’s climate mitigation commitments.
4. Policy imbalance: Focus on innovation without parallel sustainability safeguards.

How can Environmental Impact Assessment be extended to AI?

1. EIA framework: India’s EIA Notification, 2006 mandates environmental assessment for infrastructure projects.
2. Proposed extension: Inclusion of AI development and deployment within EIA scope.
3. Lifecycle evaluation: Assessment of energy use, water consumption, and emissions across AI lifespans.

What role can disclosure standards play?

1. ESG integration: Environmental impact of AI included under ESG disclosure norms.
2. SEBI alignment: Disclosure of emissions from data centres and computing activities.
3. EU precedent: Corporate Sustainability Reporting Directive (CSRD) mandates emission disclosure, including AI training.
4. Transparency outcome: Enables informed policymaking and accountability.

Which sustainable practices can mitigate AI’s impact?

1. Pre-trained models: Reduces repeated energy-intensive training.
2. Renewable energy: Powering data centres through clean energy sources.
3. Efficiency reporting: Disclosure of AI-specific environmental metrics.
4. Resource optimisation: Minimising water and energy intensity of AI infrastructure.

Conclusion

India’s AI ambitions must align with environmental sustainability. Institutionalising environmental assessment, disclosure norms, and sustainable practices is essential to prevent AI-driven ecological externalities. A regulatory framework that integrates innovation with environmental accountability will ensure AI remains a tool for inclusive and sustainable development.

PYQ Relevance

[UPSC 2023] How can Artificial Intelligence help clinical diagnosis? Do you perceive any threat to privacy of the individual in the use of AI in healthcare?

Linkage: Earlier, UPSC focused on how AI helps healthcare and affects patient privacy. Now, as AI use expands, questions are likely to include its environmental impact, especially energy- and data-intensive AI systems.