Is Artificial Intelligence a Friend or Foe to the Environment?

Is Artificial Intelligence a Friend or Foe to the Environment !

KAKALI DAS

One AI search consumes as much energy as charging your phone twice. And a single AI query uses roughly the same amount of water it takes to brew a cup of tea. Convenient as it is, AI does come with a hidden cost.

When we want to design a dream house or when we need to analyze huge data sets in minutes or we want to draft emails, enhance our CVs quickly or we want to organize our tasks and set reminders or want to look like a Ghibli art – AI is there to deliver results in no time.

Did you know that a single ChatGPT query can consume nearly 10 times more electricity than a typical Google search? In short, the ease and speed AI offers us come with an environmental cost.

Why? Because training and operating AI models require significant amounts of electricity and water, which in turn place added stress on our planet’s natural resources.

To give you some context, let me walk you through how an AI chatbot works. I type a prompt into the chatbot, which then stores my query in its database. This data is housed in servers and supercomputers located in data centers. From my phone or computer, the prompt travels to the database, where the system quickly sifts through relevant information and sends a response back to the chatbot. Finally, the AI assistant presents the answer to me.

And while those machines were crunching massive amounts of data to give me one response, they were also generating heat—because let’s face it, I’m not the only one asking questions. Millions of users around the world are feeding prompts into AI tools every day.

All this activity causes the servers and supercomputers to heat up. It’s similar to how your phone or laptop gets warm when you have too many apps or tabs open. To handle these rapid-fire queries, supercomputers rely on powerful chips called graphics processing units, or GPUs, which are designed to process huge volumes of data in very little time.

According to the United Nations Environment Programme (UNEP), the microchips powering AI rely on rare earth elements, which are frequently extracted through methods that harm the environment.

Also, the United Nations Conference on Trade and Development, in its digital economy report, highlights that producing just one 2 kg computer requires around 800 kg of raw materials.

The data centers housing the servers and supercomputers require cooling systems to prevent overheating. These centers also generate electronic waste and consume large amounts of water, which, according to UNEP, is becoming increasingly scarce in many regions.

Although water covers 71% of the Earth’s surface, less than 1% is suitable for human consumption, as the rest is either saltwater, frozen, or too deep underground to access. The small portion of freshwater that remains is also heavily utilized by industries such as agriculture and manufacturing.

Why do data centers rely on water? Primarily, to generate electricity and to use it as a liquid coolant to manage the heat produced by servers and other equipment.

According to NPR, a mid-sized data center can consume up to 1.1 million liters of water in a single day.

Several major players in the tech industry have addressed this issue. Meta, for example, has committed to reducing its environmental footprint by matching its energy consumption with renewable sources. This involves investing in new wind and solar energy projects that are integrated into local grids, including those in regions where their data centers are located.

On the water front, Meta has stated that their goal is to restore more water than they consume worldwide by 2030.

Microsoft made a similar commitment in 2020, while in 2021, Google launched a water stewardship program with the aim of replenishing 120% of the freshwater they use. In 2024, the company revealed plans to adhere to stricter environmental standards and develop a data center project in drought-prone Chile, following a temporary revocation of permission for the project by a local court.

According to Statista, there are currently over 10,332 data centers worldwide, with the US leading the pack, hosting more than 5,000. In 2015, there were only 3,600 data centers, and in just 10 years, that number has surged by approximately 7,000. This highlights the rapid expansion of data centers on a global scale.

A report from the World Economic Forum revealed that Microsoft’s carbon emissions have increased by 30% since 2020, while Google’s emissions have risen by 50%, largely due to the growth of their data center operations.

Data centers consume significant amounts of electricity to function. For instance, a single ChatGPT query is said to use nearly 10 times the electricity of a simple Google search.

Why is that? Well, an AI assistant processes vast amounts of information and generates conversational responses for users. In contrast, a Google search simply provides a list of links, leaving users to visit them and figure out the answers themselves.

For example, I asked both Google and an AI assistant the same question—“What’s the recipe for Hot & Sour Soup?” Google gave me a list of links to browse, while the AI assistant directly provided the full recipe.

Now, even Google has introduced an AI-powered feature that provides a summary of search results, in addition to offering links.

Bloomberg reported that the world’s data centers could consume as much electricity in a year as it would take to power entire countries like Italy or Australia. The growing demand for electricity driven by the surge in AI usage is now exceeding the available power supply in many regions, according to Bloomberg.

According to DW, the carbon dioxide emissions from training a single AI model are more than five times the lifetime emissions of a car.

Most of the energy consumed by data centers comes from fossil fuels like coal and gas. The burning of these fuels generates carbon emissions, further contributing to climate change.

A report from Goldman Sachs predicts that the carbon dioxide emissions from data centers could more than double between 2022 and 2030.

What we’re witnessing now is an AI revolution. Technology is rapidly advancing. Our parents didn’t have search engines to look up information instantly. Then came the internet, and Google became so ingrained in our lives that it turned into a verb. Emails replaced posts, and traditional phone calls evolved into instant texts and real-time calls.

Now, AI has entered the scene, allowing anyone to send prompts at will, anytime and anywhere.

“It’s become clear that the next generation of services requires building full general intelligence. Building the best AI assistants, AIs for creators, AIs businesses and more. That needs advances in every area of AI” – Mark Zuckerberg said

Like with any new technology, AI comes with its own set of challenges. Every year, hundreds of thousands of people die in car accidents, and we tend to accept that as part of life.

However, if even one person is killed by an autonomous vehicle, the company behind it is forced to reevaluate and redesign their system for years. In other words, our expectations for machines are much higher than for humans, making widespread adoption of AI particularly difficult. Scaling up AI adoption is a complex process, and it will continue to be a challenging endeavor.

While the conversation about AI’s contribution to pollution continues to evolve, there’s also a growing discussion about how AI can help the environment and combat climate change.

For instance, Myrose Group, a Norwegian oil spill detection system, is using AI to detect oil spills around offshore installations, helping identify if the water is contaminated.

“So we do automatic 24/7 detection that will give the operators and alert that there is probably oil spill taking place, and they will need to look at it and take their required action to cope with it” – Marius Five Aarset, CEO, Miros Group

South Korea is utilizing AI-powered bins to quickly identify and sort almost every type of plastic bottle in the country for recycling.

Scientists are also leveraging AI to track deforestation in real-time, monitor weather patterns, and improve waste management efficiency.

Experts believe that by 2030, AI could help reduce global emissions by 10%. The challenge, however, lies in finding a balance between AI’s energy usage—its contribution to pollution and its potential to combat it.

While the energy and water consumption of AI remains a concern, it will be interesting to see how companies address this in the future and whether they can effectively mitigate the impact.

Mahabahu.com is an Online Magazine with collection of premium Assamese and English articles and posts with cultural base and modern thinking.  You can send your articles to editor@mahabahu.com / editor@mahabahoo.com(For Assamese article, Unicode font is necessary) Images from different sources.