How AI Can Help and Hurt the Environment | WSJ Tech News Briefing

Summary

In this Tech News Briefing, Zoe Thomas discusses the dual impact of AI on climate change. AI has the potential to reduce climate change by improving efficiency in sectors like aviation and flood forecasting. However, it also contributes significantly to greenhouse gas emissions and water consumption, with models like Bloom and GPT-3 having substantial carbon footprints. The discussion highlights the importance of considering AI's location and integration to minimize its environmental impact, with companies like Google aiming to replenish water usage and suggesting more mindful AI deployment.

Takeaways

• 🌐 Technology, particularly data centers, has a significant impact on the climate, emitting greenhouse gases comparable to the aviation industry and using substantial amounts of water.
• 🤖 Artificial Intelligence (AI) can be utilized to improve efficiency and reduce climate change impacts, such as by optimizing flight paths and predicting floods.
• 🛫 Alphabet's Google and American Airlines have used AI to reduce the creation of vapor trails in planes, which contribute to global warming, and to forecast river floods.
• 💧 AI startups, especially in San Francisco, are simplifying the process for companies to access clean power, which is also being adopted by everyday people.
• 🖼️ Image generators are using AI to depict what a warmer world might look like, including scenarios like ocean encroachment and wildfires.
• 🔍 The environmental impact of AI itself is a growing area of research, with studies beginning to assess the lifetime carbon footprint of machine learning models.
• 🌿 A study by AI app developer Hugging Face revealed that the carbon footprint of the Bloom model, with 176 billion parameters, was so large that it doubled the total emissions of the model.
• 🔧 Manufacturing hardware like GPUs, which are used in deep machine learning, involves a significant amount of pure water and rare metals, adding to the climate cost.
• ⚡️ OpenAI's GPT-3 model had a carbon footprint over 20 times higher than Bloom and consumed three times as much power, highlighting the scale of energy use in AI.
• 💧 AI models like GPT-3 require water for cooling, with estimates suggesting a basic conversation could use up to a 500 milliliter bottle of water.
• 🌊 Google's large language model Lambda used around a million liters of water for training, and the company's data center water consumption increased by 20% in 2022.
• 🌳 Google has a 2030 target to replenish 120% of the water it consumes and is focusing on using non-fresh water sources like wastewater or seawater.
• 📍 The location of AI model training can significantly affect carbon emissions, with states like California having more renewable energy sources than Virginia, which relies more on fossil fuels.
• 🌍 Internationally, Microsoft noted that its Asian data centers had three times worse water use effectiveness compared to those in the U.S., due to warmer climates and different energy sources.
• 🔄 To reduce AI's climate impact, one practical step is to avoid integrating AI into areas where it is not necessary, as this can increase the climate cost of everyday actions.

Q & A

What is the potential impact of technology on climate change?

Technology, particularly data centers, can have a significant impact on climate change by emitting greenhouse gases comparable to the aviation industry and consuming large amounts of water. However, artificial intelligence (AI) can also help mitigate these effects by improving efficiency and understanding climate impacts better.

How can AI be utilized to reduce the impact of climate change?

AI can improve efficiency in various sectors. For example, Google and American Airlines have used AI to reduce vapor trails in planes, which contribute to global warming. AI is also used to forecast river floods and recommend eco-friendly routes, and startups are using it to simplify the process of obtaining clean power.

What are some of the ways AI is being used in everyday life to address climate change?

AI is being used in image generators to show what warmer worlds might look like, including ocean encroachment and the appearance of the world on fire if temperatures rise significantly. This helps to raise awareness and potentially drive action on climate change.

What studies have been conducted on the energy use of AI and its impact on climate change?

Studies by organizations such as Hugging Face have explored the lifetime carbon footprint of machine learning models, revealing that factors beyond just the energy used in training, such as manufacturing hardware, can significantly contribute to a model's total emissions.

How does the carbon footprint of the AI model Bloom compare to other models?

Bloom, a machine learning model with 176 billion parameters, was found to have a carbon footprint that doubled when considering factors outside of just the energy used in training. This includes the manufacturing of GPUs and other hardware.

What is the scale of energy use for models like OpenAI's GPT-3 during training?

OpenAI's GPT-3 has a significantly higher carbon footprint, more than 20 times higher than Bloom, and consumes three times as much power. It also uses enough energy over its training period to power an average American home for over four decades.

How does the water use of AI models compare to traditional data centers?

Research indicates that AI models like GPT-3 require a considerable amount of water for cooling, with GPT-3 needing the equivalent of a 500 milliliter bottle of water for a basic conversation. Google's large language model, Lambda, used around a million liters of water for training alone.

What steps is Google taking to address its water consumption in data centers?

Google is attempting to use non-fresh water sources such as waste water, industrial water, or seawater for cooling in their data centers. They have also set a 2030 target to replenish 120% of the water they consume.

What are some practical steps that can be taken to reduce the climate impact of AI?

One practical step is to use AI less when it is not necessary, such as in search engines where existing software works efficiently without AI integration. This can help reduce the climate cost of basic actions that have been performed efficiently without AI.

How does the location of AI models affect their carbon emissions?

The location can significantly impact carbon emissions because the energy sources used in different states or countries vary. For instance, training models in California, where there is a lot of wind power, can result in lower emissions compared to states that rely on fossil fuels.

What is the significance of the water use effectiveness of Microsoft's Asian data centers compared to their U.S. locations?

Microsoft's Asian data centers were found to have three times worse water use effectiveness than their U.S. locations, indicating that the geographical location of AI models can have a substantial impact on water consumption and overall environmental impact.

Outlines

00:00

🌐 AI's Role in Climate Change Mitigation

The first paragraph of the Tech News Briefing discusses the significant impact technology, particularly data centers and artificial intelligence (AI), has on climate change. It highlights that data centers emit greenhouse gases equivalent to the aviation industry and consume substantial water resources. However, AI has the potential to mitigate these effects by improving efficiency in various sectors. For instance, Google and American Airlines have utilized AI to reduce vapor trails that contribute to global warming. AI is also being used to forecast floods and recommend eco-friendly routes. Startups are leveraging AI to facilitate the transition to clean energy. Additionally, AI-powered image generators are visualizing the effects of climate change, such as ocean encroachment and wildfires, to raise awareness.

05:01

💧 The Environmental Footprint of AI Development

The second paragraph delves into the environmental implications of AI's own development and operation. Recent studies have begun to assess the energy consumption and carbon footprint of AI models. One study by Hugging Face examined the lifetime carbon footprint of the Bloom AI model, revealing that factors beyond training, such as manufacturing hardware like GPUs, significantly contribute to the model's overall emissions. The energy and water consumption during the training of large AI models like OpenAI's GPT-3 and Google's Lambda are staggering, with GPT-3 consuming enough energy for an American home for over 40 years and Lambda using a million liters of water in training alone. The paragraph also touches on Google's efforts to use non-fresh water sources and its commitment to water replenishment by 2030. It concludes by suggesting ways to reduce AI's climate impact, such as integrating AI judiciously and considering the environmental cost of training AI models in different geographic locations.

Keywords

💡Climate Impact

Climate impact refers to the effects that human activities have on the environment, particularly in terms of global warming and the emission of greenhouse gases. In the video, it is discussed how technology, specifically AI, can both contribute to climate change and also be used to mitigate its effects, such as by improving the efficiency of data centers and reducing the environmental footprint of industries like aviation.

💡Artificial Intelligence (AI)

Artificial Intelligence is the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions. The script highlights AI's dual role: it can exacerbate climate change through high energy consumption but also help combat it by enhancing efficiency in various sectors, such as aviation and energy management.

💡Data Centers

Data centers are large repositories of data that require significant amounts of energy and water for their operation, contributing to greenhouse gas emissions. The script points out that AI can help optimize data center operations to reduce their environmental impact, but it also emphasizes the need to address AI's own climate impact from data centers.

💡Greenhouse Gas

Greenhouse gases are gases in the atmosphere that trap heat, leading to the greenhouse effect and global warming. The script mentions that data centers emit about the same amount of greenhouse gases as the aviation industry, underscoring the need for AI to help reduce these emissions.

💡Efficiency

Efficiency in this context refers to the optimal use of resources to achieve the maximum output with the minimum input. The video discusses how AI can improve efficiency in various sectors, such as by helping planes create fewer vapor trails that contribute to global warming, thus reducing the climate impact.

💡Alphabet's Google

Alphabet's Google is used as an example in the script to illustrate how AI has been used to improve environmental outcomes. Google has employed AI to help planes create fewer vapor trails, which are known to contribute to global warming, showcasing the practical application of AI in reducing climate change impact.

💡Machine Learning Model

A machine learning model is a type of AI that improves its performance on a specific task through experience without being explicitly programmed for it. The script discusses the carbon footprint of a machine learning model named Bloom, developed by Hugging Face, to highlight the environmental impact of AI development and usage.

💡GPU (Graphics Processing Unit)

A GPU is a specialized electronic circuit designed to rapidly manipulate and alter memory to accelerate the creation of images in a frame buffer intended for output to a display device. The script mentions the manufacturing and use of GPUs in deep machine learning as part of the climate cost, emphasizing the energy and materials used in their production.

💡Water Use

Water use in the context of the script refers to the significant amount of water required for cooling data centers. The video discusses the water consumption of AI models like Chat GPT3 and Google's Lambda, indicating that AI's environmental impact extends beyond energy use to include water resources as well.

💡Sustainability

Sustainability generally refers to the ability to maintain processes or states in a certain way over the long term. In the script, sustainability is related to the efforts to reduce the environmental impact of AI and data centers, such as Google's 2030 target to replenish 120% of the water it consumes.

💡Eco-friendly Routes

Eco-friendly routes are paths or methods of travel that have a lower environmental impact. The script mentions that AI can be used to recommend such routes, which is an example of how technology can be leveraged to promote sustainable practices and reduce the carbon footprint of transportation.

💡AI's Own Climate Impact

This phrase refers to the environmental effects caused by the operation and development of AI systems themselves. The script discusses studies that have begun to measure the carbon emissions and energy consumption associated with training and running AI models, indicating a growing awareness of the need to make AI more sustainable.

Highlights

Technology, particularly data centers, contributes significantly to greenhouse gas emissions, comparable to the aviation industry, and consumes a lot of water.

Artificial intelligence (AI) can help mitigate the climate impact by improving efficiency in various ways.

Alphabet's Google and American Airlines have utilized AI to reduce vapor trails in planes, which contribute to global warming.

AI is also being used to forecast river floods and recommend eco-friendly routes.

Startups, especially in San Francisco, are employing AI to simplify the process for companies to obtain clean power.

AI-powered image generators are visualizing the effects of a warmer world, such as ocean encroachment and wildfires.

AI's own climate impact is a growing area of research, with studies beginning to assess its energy use and carbon footprint.

Hugging Face's study on the AI model Bloom revealed that factors beyond training, like manufacturing hardware, significantly contribute to the model's lifetime carbon footprint.

The manufacturing of GPUs for deep machine learning involves substantial use of pure water and rare metals, adding to the climate cost.

Open AI's GPT-3 model has a carbon footprint over 20 times higher than Bloom and consumes three times as much power.

Bloom's energy use for training is equivalent to powering an average American home for over 40 years.

A single training run of Bloom has 25 times the emissions of a round-trip flight from New York to San Francisco for one passenger.

Research from the University of California Riverside indicates that AI models like Chat GPT-3 require substantial water usage for cooling.

Google's large language model Lambda used around a million liters of water just for training.

Google has a 2030 target to replenish 120% of the water it consumes and uses non-fresh water sources when possible.

Reducing AI's climate impact can be achieved by using it judiciously and avoiding unnecessary integration into systems that already function well without it.

The location of AI model training can significantly impact carbon emissions due to differences in energy sources between states and countries.

Microsoft found that its Asian data centers' water use effectiveness was three times worse than in the U.S locations.

Transcripts

[Music]

welcome to Tech news briefing it's

Monday October 2nd I'm Zoe Thomas for

The Wall Street Journal

technology can have a big impact on the

climate data centers admit about the

same greenhouse gas as the aviation

industry and consume a lot of water but

artificial intelligence can help us

better understand and cut down on that

if we can limit ai's own climate impact

here to talk about how to do that is

nuha Dolby who reported on this for the

wsj's sustainability pro team Newhall we

talk a lot on this show about the

potential of AI one thing it could do is

help reduce climate change impact how

could it do that yeah so there are a

number of ways but broadly it's about

improving efficiency so for instance

alphabets Google and American Airlines

have used artificial intelligence and

they've used that to help planes create

fewer Vapor Trails and those Vapor

Trails actually contribute to global

warming the companies also use it to

forecast river floods and they can use

it to recommend eco-friendly routes

there are startups too based out of

predominantly San Francisco

one's using AI to simplify the process

for companies to get clean power and

that's also being used by Everyday

People too so lots of people and image

generators that have become pretty

pretty popular have used it to generate

images of what warmer worlds will look

like so things like ocean encroachment

to looking at what the world would look

like on fire if it gets hot all right

let's talk about ai's own climate impact

have there been any studies about Ai and

its energy use yeah so this is kind of a

newer area that people are looking into

but there are a couple studies one came

out of hugging face which is an AI app

developer and the research scientist

there had decided to map the lifetime

carbon footprint of a machine learning

model with 176 billion parameters and

that model is called Bloom

so the factors outside of the energy

used just in training the model is

something that a lot of research in the

area hasn't really factored in wound up

being so large that they actually

doubled the total emissions of the

entire model so for instance

manufacturing a GPU or like a graphics

Processing Unit it's a piece of Hardware

that's in most computers and is also

used in this deep machine learning

manufacturing those involves a lot of

pure water and rare metals and that kind

of thing and that'll add to the climate

cost Bloom just alone used more of a

thousand of those gpus and that's just

one of many factors that this research

took into account so more impact than

just its energy use but if we were to

look at Bloom's energy use do we have a

sense of the scale there so for a model

of similar size that might be more

familiar open AIS chat GPT 3 had

significantly higher carbon emissions so

more than 20 times higher and it

consumed three times as much power as

bloom bloom also used enough energy she

ingested training over a number of

months to power the average American

home for just over four decades and the

training run also had 25 times the

emissions of just one passenger's

round-trip flight from New York to San

Francisco so if you're feeling guilty

about that transatlantic flight you took

recently here's some context for it what

about water use data centers typically

use a huge amount of water to cool

themselves what is the water use like

for AI so research out of the University

of California Riverside has shown that

about chat gpt3 needs to drink like a

500 milliliter bottle of water for just

a basic conversation of between 20 to 50

inquiries depending on where that

electricity is generated gpt4 probably

uses more the research out of the

University also did estimates for

Google's large language model known as

Lambda and that one used around a

million liters of water for its training

alone Google's on-site data center water

consumption overall in 2022 has also

gone up by around 20 compared with the

year before have the company said

anything about ways that they might

reduce that Google has said that when

they try to use water they try to use

things that aren't fresh water so things

like waste water industrial or even sea

water when Google is a 2030 Target to

replenish 120 of the water it consumes

are there ways to reduce ai's climate

impact things that companies or users

might be able to do one thing people can

do of course is is using less but a

practical step in limiting those

emissions is just to not integrate AI

into things that it doesn't need to be

in our search engines work now we have

lots of software that works just fine

and as companies try to integrate AI

into it they're increasing the climate

cost of all those basic actions that

everyone does and has done just fine

before and can continue to do just fine

without that integration what about in

terms of location for example should we

be thinking about where these AI models

are housed in the U.S for instance where

there's no Central Electric grid

training models in one state versus

another can have a pretty big impact on

carbon emissions and you don't actually

have to move to do that because all this

is done over the internet

in California where we have a lot of

wind power there's a good shot that the

energy is producing less emissions and

if you're using it in Virginia which has

lots of coal and other fossil fuels

internationally because it's typically

warmer in Asia Microsoft said last year

that it's Asian data centers actual

water use Effectiveness was three times

worse than that of their locations in

the U.S so if you're training an AI

model you could triple your water use

just by having that be in Asia all right

that was nuha Dolby who reported on this

for wsj pro sustainability

and that's it for Tech news briefing

Today's show was produced by Anthony

bansi with supervising producer Melanie

Roy I'm Zoe Thomas for The Wall Street

Journal we'll be back tomorrow thanks

for listening

[Music]

[Music]