India must focus on AI and its environmental impact

Using Synthetic Intelligence (AI) takes up a lot dialogue in sectors that vary from well being care to agriculture. However issues about its impression on the setting haven’t attracted a lot dialogue. Based on an OECD working paper, “Measuring the Environmental Impacts of Synthetic Intelligence Compute and Functions”, the event of AI algorithms comes with sure environmental prices reminiscent of an elevated carbon footprint which exacerbates local weather change-related challenges. The report says that the worldwide ICT trade (together with {hardware} reminiscent of televisions) is estimated to be accountable for 1.8%-2.8% of worldwide greenhouse fuel (GHG) emissions; different calculations have a determine as excessive as 2.1%-3.9%.

It’s price noting that knowledge in regards to the carbon footprint of AI fashions and their use will not be at all times genuine. A Google report (August 2025) claims {that a} single textual content AI immediate consumes electrical energy of solely 0.24 watt-hours. What could appear to be low ranges of electrical energy consumption has additionally drawn criticism for the report’s incomplete and deceptive conclusions.

Impression on the setting

In September 2024, a difficulty be aware by the United Nations Surroundings Programme (UNEP), on the environmental impression of the complete AI life cycle, stated that home AI servers could utilise 4.2 billion cubic meters (bcm) to six.6 bcm of water in 2027, resulting in water shortage. The be aware additionally refers to a examine which signifies that coaching a single Massive Language Mannequin (LLM) can generate nearly 3,00,000 kilograms of carbon emissions.

Equally, a examine, “Vitality and Coverage Concerns for Deep Studying in NLP” (June 2019), has assessed that the coaching of a single massive AI mannequin entails the emission of over 6,26,000 kilos of carbon dioxide. That is equal to emissions attributable to 5 automobiles of their lifespan. A take a look at the power consumption in using ChatGPT, a extensively used AI digital Assistant, helps perceive how AI algorithms impression the setting and contribute in the direction of local weather change. Based on a examine, “Navigating New Horizons” (UNEP, July 15, 2024), any request made by ChatGPT results in power consumption that’s 10 occasions greater than what it could be by a Google search.

Options, the world versus India

In 2021, UNESCO launched its “Suggestion on the Ethics of Synthetic Intelligence”, which emphasised recognising the “detrimental impacts of AI on societies, setting”. These non-binding suggestions have been adopted by round 190 nations. America and the European Union (EU) seem like the outstanding jurisdictions which have proposed laws coping with the environmental impression of AI, because the Synthetic Intelligence Environmental Impacts Act of 2024 and The European Union’s decision on harmonized AI guidelines.

Since there are international conversations in regards to the carbon value of AI use and deployment, India additionally must recognise the environmental prices of creating AI fashions. Present discussions now on AI and local weather change are on how AI can assist defend the setting, however with out going into the demerits of creating massive AI algorithms.

Step one to deal with this problem is to hold out an train to measure the environmental impacts of creating and deploying AI fashions. In India, an Environmental Impression Evaluation (EIA) is necessary as within the EIA Notification, 2006. Whereas an EIA is commonly performed to guage initiatives in regards to the setting reminiscent of river water initiatives, its scope will be prolonged to incorporate assessing the impression of improvement of AI algorithms on the setting.

The federal government may additionally concentrate on the institution of measuring requirements to be able to assess AI’s impression. It may be finished by involving stakeholders reminiscent of tech firms which are creating large-scale AI algorithms, assume tanks, and non-governmental organisations which are working towards decreasing carbon footprints and mitigating different environment-related challenges. It will assist in constructing consensus on terminology, requirements, and constant indicators and reporting necessities, finally resulting in knowledgeable coverage selections.

One other important train is knowledge assortment which will be finished by deploying sustainability metrics reminiscent of GHG emissions, power, water and pure assets consumption which are utilised by AI algorithms. Environmental prices past power, such because the impacts on freshwater and land use, is also evaluated.

As part of disclosure requirements

The federal government may discover the opportunity of making the environmental impression of creating and deploying AI fashions as a part of environmental, social and governance (ESG) disclosure requirements by the Ministry of Company Affairs and Safety and the Securities and Change Board of India. It could actually take inspiration from the EU, as its Company Sustainability Reporting Directive (CSRD) framework requires the disclosure of emissions knowledge from knowledge centres and excessive compute actions, which incorporates coaching of LLMs.

The main target ought to shift to together with AI as part of options towards international sustainability targets. There are a number of AI sustainable practices that may be adopted to mitigate the antagonistic impression of AI on the setting reminiscent of deploying pre-trained fashions, utilizing renewable assets to energy knowledge centres, and reporting AI-specific estimates.

Amar Patnaik is a lawyer and founding accomplice of A&N Authorized Options LLP, a former CAG bureaucrat and a former Member of Parliament within the Rajya Sabha.

Printed – January 14, 2026 12:08 am IST