AI and ESG: A ChatGPT Prompt May Use Half a Litre of Drinking Water

• Though AI has a huge environmental footprint, it is also used to detect emissions and optimise supply chains.
• The data centre is the elephant in the room; some consume as much water daily as 1,000 farms or homes.
• Experts say making AI more efficient could end up being counterproductive.

Generative AI tools like ChatGPT and Gemini are quite literally like a friend in need—one who comes with a repository of knowledge, suggestions, and advice at your fingertips. But as their use has expanded dramatically, checks and balances are at stake, particularly regarding their environmental impact.

Sustainability-focused investors are increasingly worried about the environmental, social, and governance (ESG) impact of these AI tools and are calling them out, as they require a humongous amount of data and computing power to operate, as well as large amounts of electricity and water.

Many technology companies failed to take pre-emptive action and did not expect artificial intelligence (AI) to grow on an epic scale. As a result, they are grappling with their ESG commitments, and some have even been forced to put their net-zero goals on ice.

READ MORE: UN: AI Fuelled 150% Rise in Tech Giants’ Data Centre Emissions

AI Water Use: A Rising Concern

Sondre Myge, a sustainability expert who is now Head of ESG at Skagen Funds, claims that a single ChatGPT prompt can use approximately half a litre of drinking water (a figure he got from ChatGPT itself). Other recently conducted studies, eerily similar to this statement, point out that water is predominantly used to cool data centres and run AI efficiently. To cap it all, data centres, to meet purity standards needed for equipment, hinge on clean, potable water.

Consequently, this could lead to competition between tech companies and local communities where water is scarce. Not just that, if companies move their data centres to unaffected areas, it could even create water stress in places where it did not exist before.

Energy Use: Another Worry, But Solutions Exist

Surprisingly, AI data centres are consuming more electricity than the entire country of France. Current air-cooling systems are becoming insufficient to cool down the intense heat AI generates, and replacing them are liquid cooling systems, which are 1,000 times more efficient than the former.

This demands infrastructure redesign for the smooth flow and reuse of cooling liquids. Norway is making the most of this by using excess heat from data centres to warm buildings, which is a sustainable solution.

ALSO READ: As AI Grows, Data Centres Face Soaring Energy Demand

According to experts, things are turning the corner. With closed-loop cooling systems, newer data centres recycle water and reduce water waste. They opine that although generative AI is energy- and water-intensive, the repercussions in the form of environmental costs are not as high as feared in modern facilities.

Despite worries, some quarters believe that, for the present, AI's impact is manageable. Technology companies are increasingly using electricity from renewable sources like wind and solar. This signals that the tech industry, in its own capacity, is reducing the environmental impact—even though it did not anticipate AI's unprecedented growth.

Hortense Bioy, head of sustainable investing research at Morningstar, said that long-term sustainability still remains a question mark. She substantiates her claim by referring to the Jevons Paradox, which suggests that greater efficiency leads to increased use of resources rather than a decrease. AI is primed for wider use—and at a low cost—which drives energy and resource consumption.

While AI has a considerable environmental footprint, it also offers environmental benefits. For example, AI is used in supply chain optimisation, streamlines production, and has even been reported to detect methane leaks—a powerful greenhouse gas. Besides, tech giant Google uses AI to monitor the energy needs of its data centres.

ALSO READ: Microsoft Integrates AI for ESG Reporting

A report by the International Energy Agency notes that, if used appropriately and correctly, AI could ultimately reduce more energy than it consumes.

So, the takeaway is that AI is a double-edged sword. Its applications range from reducing overproduction to managing waste and cutting emissions, but it could also be detrimental if not used thoughtfully. Above all, AI development should align with global sustainability goals to maintain proper checks and balances.

Ends/

Looking for top sustainability and AI specialists to scale your business operations? Our broad network of ESG and sustainability professionals is ready to support you with a diverse range of services.

Stay updated with KnowESG’s Technology News for the latest insights and analysis.

Explore our recent ESG Event announcements.

Source: Citywire Selector