ChatGPT is Thirsty

ChatGPT is really, really thirsty!

A reader submitted an article this week which I found very interesting. It has to do with water and Artificial Intelligence (AI). 

But in this article, AI isn’t the solution, it’s part of a giant problem.

The simple truth is that we take water for granted. That’s because, in most of the U.S., when we turn on the tap, clean water comes out. Because the infrastructure that carries that water is underground, we don’t ever think of it.

The sad truth about the American water infrastructure is that it’s old. Some of it was installed over 100 years ago. And much of it needs to be replaced. But that’s a huge problem. Fixing the water infrastructure will cost a trillion dollars.

The U.S. government committed billions of dollars to solving this problem, with more to come by 2025. That much money means a lot of companies will compete to come up with solutions to America’s water supply issues.

But there’s a much more modern problem.

That article pointed out that AI needs a lot of water.

The computing resource load needed to fuel Artificial Intelligence and generative tools like ChatGPT is enormous. So much that in the latest Microsoft environmental report their global water consumption spiked 34% from 2021 to 2022 which was enough to fill more than 2,500 Olympic-sized swimming pools. And researchers tie it to AI computing.

“A.I. tools fueled a 34% spike in Microsoft’s water consumption, and one city with its data centers is concerned about the effect on residential supply”

With new AI programs launching nearly weekly this will continue to get worse. However, we can also look to AI to help find solutions.

Demands on both our water and electrical infrastructure will continue to grow. AI technologies will provide utilities with the tools to optimize distribution, improve efficiency, enhance sustainability, and respond to emerging challenges effectively.

As the world grapples with a growing population and the uncertainties of climate change, AI is becoming indispensable allies in safeguarding our most precious resource: clean and accessible water. Harnessing the power of AI and language models is not just an option; it’s a necessity.

It is all about balance.

In our rapidly advancing world, technology has become an integral part of our daily lives. From AI and beyond, technology has reshaped the way we live, work, and even think. While these innovations bring numerous benefits and opportunities, it is crucial to recognize that technology is a double-edged sword. The key to navigating this digital age successfully lies in achieving a balance between the advantages and challenges that technology presents.

I did not want to end this article without letting ChatGPT defend itself for being a water hog!

So here you go:

ChatGPT what do you have to say for yourself on using so much water?

ChatGPT: 

The Hidden Costs of Artificial Intelligence and GPT-3.5: How Computer Resource Needs Impact Water Supply

Introduction

Artificial Intelligence (AI) and advanced language models like GPT-3.5 have ushered in a new era of technological progress, offering transformative capabilities across various industries. However, the exponential growth in AI applications has led to a significant and often overlooked consequence—the staggering demand for computer resources, which can have profound implications for our water supply. In this article, we explore how AI and GPT-3.5’s resource requirements are affecting water availability and the need for sustainable strategies to mitigate this impact.

The Data Dilemma

AI and GPT-3.5 thrive on vast datasets, and training these models requires massive computing power. The processing of such data consumes substantial amounts of electricity. Data centers, where much of this computation takes place, require immense cooling systems that use water to dissipate heat. In regions with water scarcity, this additional demand on water resources can strain local water supplies.

Energy Intensity

AI workloads are notorious for their energy consumption. Training a single AI model can consume as much energy as several households use in a year. Power plants that generate this energy often rely on water for cooling purposes, resulting in a nexus between AI’s energy appetite and water use. Consequently, regions facing both energy and water scarcity may find themselves in a precarious situation.

The Impact on Cooling Systems

Data centers housing AI infrastructure often rely on water-intensive cooling systems to maintain optimal operating temperatures. These cooling systems can consume vast quantities of water, particularly in areas where water resources are already under stress. This increased demand for cooling water can exacerbate local water scarcity issues.

Infrastructure Strain

As organizations expand their AI capabilities, they often build new data centers or expand existing ones to accommodate the growing computational demands. This can strain local water infrastructure and lead to competition for water resources among industrial, agricultural, and residential users.

Regional Disparities

The impact of AI and GPT-3.5 on water resources is not uniform. Regions with abundant water may not experience immediate issues, while those already grappling with water scarcity face exacerbated challenges. This can lead to inequities in access to water resources and potentially hinder economic development.

Conclusion

Artificial Intelligence and advanced language models like GPT-3.5 have the potential to revolutionize industries and improve our lives. However, their voracious appetite for computer resources, particularly water and energy, poses significant challenges. As the use of AI continues to expand, it is imperative that we develop and implement sustainable strategies to mitigate its impact on water supplies. Balancing technological advancement with responsible resource management is crucial to ensure a sustainable and equitable future for all.

For the Good,

Michael Nichols and ChatGPT

Numbers to Know

4000 BC

Archaeologists discovered the first copper water pipes in palace ruins of India’s Indus River Valley dating back to 4000-3000 B.C. (Tomorrows World Today)

700

It is estimated that an average of 700 water mains will break in cities and towns across North America per day. (Water World)

2 Billion

In 2021, over 2 billion people live in water-stressed countries. (World Health Organization)

What’s New in Sustainable Investing

California Poised To Enact Climate Disclosure Laws. What That Means for Investors

The California State Assembly passed two sweeping bills addressing climate change this week – one that requires big companies that do business in California to disclose their greenhouse gas emissions, and a second that would require companies to report every year on their climate-related financial risk. (Morning Star)

Biden-Harris Administration Launches New Large-Scale Water Recycling Program with $180 Million from Bipartisan Infrastructure Law

The Department of the Interior announced the launch of a new large-scale water recycling program and made $180 million in initial funding from the Bipartisan Infrastructure Law available for projects aimed at creating new water supplies. (U.S. Dept of Interior)

Video Of The Week