Shadowing Practice: How much energy do AI searches really use? - The Climate Question podcast, BBC World Service - Learn English Speaking with YouTube

Shadowing Controls

0% completed (0/206 sentences)

I think this is 2025 data.

⏸ Paused

Speed:

Repeat Count:

Wait Mode:

Sub Sync:0ms

All Sentences206 sentences

I think this is 2025 data.

AI consumed about 0.5, half of 1% of the world's electricity.

And you've got to remember electricity is a subset of total energy use.

So it's more like kind of 0.1, 0.2% of total energy use.

Hello and welcome to The Climate Question.

This week we're answering your comments and questions, including what's the carbon footprint of streaming TV?

And are you a climate optimist or pessimist?

From the BBC World Service, I'm Greya Jackson.

Now, let me introduce you to our panel.

We have Akshat Rati, Senior Climate Reporter at Bloomberg Green.

Welcome.

Hi, Greya.

We have Caroline Steele, host of CrowdScience on the BBC World Service.

Welcome, Caroline.

Hello.

Thanks for having me.

And the BBC's climate editor, Justin Rowlett.

Hi there.

Thank you very much for having me.

Now, let's have our first question from Albert.

He wrote, the carbon footprint of electronic devices is significant and growing, encompassing emissions from manufacturing, usage and disposal.

This was not included in your recent programme, considering how individuals could lower their carbon footprint.

I particularly abstain from frivolous phone camera and social media use, along with downloading films from Netflix.

Now, Albert also talked about the carbon cost of AI and data centres, so we'll come to that as well.

But I just wanted to say that typically in the West, the biggest part of someone's carbon footprint is transport, whether they fly, whether they drive a petrol car, electric car, heating and cooling.

Their home is also a big part of their carbon footprint.

And if they eat meat or dairy, then diet too.

And that is why we focus on it, because those are the most meaningful ways in which people can reduce their carbon footprint.

But I do really want to talk about this issue of digital carbon footprints, especially when it comes to streaming.

So Akshat, what is the carbon footprint of streaming, say, an hour of TV or scrolling on social media?

This is the part of the question that to me is a little bit difficult to answer because there is no simple way of calculating your specific carbon footprint of a particular digital usage.

So just to try and figure out what the range could be, here's one example from a 2020 study.

If you had a five-hour Zoom call, your carbon footprint, according to a 2020 study, could be between four kilograms of carbon dioxide to 215 kilograms of carbon dioxide.

Wow, that's a massive difference.

Because where you get your electricity from varies.

What type of computer you're using varies.

How far your internet has to travel to you, etc. varies.

there is a huge difference in those numbers.

But at the end of the day, to me, the amount is actually quite small.

So again, trying to look at the total impact that internet usage has on the world, it is between three and 5% of CO2 emissions if you account for everything.

So you are accounting for the emissions of making the electronic products to running them on the internet, etc,

etc. So it is important that if you're thinking about a personal carbon footprint, those three things that you put in, fly less, eat less meat, and be conscious whenever you can be, is the best way to think about your carbon footprint.

So Netflix released some data in 2020 and they said it was under 100 grams of CO2 per hour streamed.

And that's similar to driving a petrol car, 400 meters.

iPlayer published some statistics in 2016, similar, 98 grams of CO2.

But again, Purdue University put it up at 440 grams of CO2 per hour streaming.

So like you say, the estimates really, really vary.

Caroline, can we look at the carbon footprint of devices, making them and then disposing of them?

Yeah, so I think it's useful to think about streaming, but a huge part of the carbon footprint of devices comes from the production.

So I found an interesting report.

It's a digital economy report published by the UN Trade and Development in 2024 that basically breaks down different devices and their carbon footprint in making them and using them.

And interestingly, battery-powered devices like phones or tablets, the vast majority of their carbon footprint comes from the production of the device.

So that's things like extracting the materials needed, shipping the parts all over the world, delivering them to the consumer.

So making a smartphone produces about 50 kilograms of carbon dioxide equivalent, which is about the same as driving from Berlin to Prague.

That's before you've even turned your phone on.

And then in the lifetime of a phone and using the phone, you're likely to only produce about eight kilograms of carbon dioxide equivalent.

So, you know, that's the vast majority of the carbon footprint there comes from making the device.

And then for sort of mains powered things like TVs or desktops, there's still a significant chunk in the production, but it's not quite as big.

So for a TV, producing a TV releases about 200 kilograms of carbon dioxide equivalent.

But then when you use a TV, you're releasing about 600 kilograms of carbon dioxide equivalent.

So there it's slightly more shifted towards using the device.

But I guess this all goes to say that basically something we can do that really makes a difference is keep using our devices for as long as we can, which I know is hard when things like phones aren't really built to last.

And it's tempting to upgrade and get the best camera and that kind of thing.

It's hard to fix them as well when they break, although there are some sort of interesting ideas around legislation and how we might reuse our devices more.

I mean, Justin, AI was also featured in Albert's question and data centres.

And that's something that you've been looking at.

Yeah, I mean, I think it comes back to what Akshat said, that actually it's quite a small proportion of most people's carbon footprint.

And we, for a long time, the big AI companies didn't really release data, but increasingly they're releasing data.

And it shows that actually the carbon footprint of an individual search is actually extremely low.

And even if you're doing 10 or even 100, it's a tiny proportion compared to other things that you do.

Like, I mean, anything that involves heating stuff up typically takes a hell of a lot of energy.

So boiling a kettle or doing the washing would be much higher than, for example, using AI.

And the issue with AI, it's quite interesting, the issue with AI is more that the footprint of big data centres is huge and it's where they are.

So some countries disproportionately are building these data centres.

For example, in America, it's a huge proportion of additional electricity use, partly actually because America doesn't have a very rapid increase in electricity demand because it's not investing heavily in electric cars and stuff like that.

So there isn't an underlying demand.

So a larger proportion of it's in data centers and obviously the big AI companies are there in America.

But so so it may be an issue locally in America, but it's not a huge issue in terms of people's carbon footprints.

I was just looking at some studies on it.

If you I think this is 2025 data data, AI consumed about 0.5, half of one percent of the world's electricity.

and you've got to remember electricity is a subset of total energy use so it's more like kind of 0.1 0.2 percent of total energy use

so it fits very firmly um in Akshat's point about you know go for the big stuff you know if you're worried about your carbon footprint probably start focusing on the big stuff first of all and then kind of start to worry about how many

AI searches you're doing worrying about things like you know the carbon footprint of activities yeah yeah and I guess the other thing to point out is these

devices have a huge number of benefits right they enabled us to access information that we never would have been able to you know 20 years ago

ai has lots of benefits you know when you talk to a i spoke to nvidia about this and they had like they did real off fantastic

examples of you know ai used to sort sort um you know rubbish in in in rubbish centers to separate it out you know use cameras and they identify

i mean all sorts of things predict you know really fine uh weather forecasts that allowed you to turn off power generation from fossil

fuel sources to balance off you know in anticipation of the wind picking up and there being more kind of wind power for example so all sorts of applications i

mean i don't think anyone's done a good peer-reviewed article that looked at the impacts but clearly there are a balance on either side although i suspect at the moment that there's a kind of net draw of additional energy from ai than a contribution from it just because it's quite a novel technology.

Caroline?

I have a question.

Oh!

Are you writing?

Just click on the top of the screen.

Comment below or email the client.

I thought I'd use one.

We'll address it in a future program.

Go on.

So is AI significantly worse than doing like a normal search engine search?

100

So if you just say Google something, is that better than asking AI?

101

Yes.

102

Why is that?

103

Google doesn't automatically serve you an AI answer, which it does.

104

It does now.

105

So about 10 times is a good rough number to know between a normal Google search, searching for a Wikipedia page or doing a question search.

106

And the problem, the issue with AI, obviously the degree, the kind of questions you ask from a simple, I mean, lots of people use AI to effectively do the kind of search that you do on Google any other search engines, lots of other search engines available by the way.

107

But if you do a really complex one, if you're asking for detailed analysis or even more so if you are agentic AI,

108

if you're asking the AI to go off and do stuff for you and maybe commission additional kind of searches or you know or AI activities then obviously there's a huge multiplication of the amount of effort it takes.

109

A hundred thousand, like if you're asking it to book a flight it could cost a thousand times one google search.

110

Wow.

111

But the other thing that they do, so they're trying to rationalize it.

112

So if it's a familiar search, then it's kind of bounded.

113

They don't do a huge kind of scrape of the internet.

114

They kind of know roughly where the good sources are.

115

And so it will use those.

116

And then if you demand a deeper search, obviously that takes considerably more kind of energy and activity.

117

To me, any activity that uses electricity is getting cleaner anywhere in the world.

118

And so your carbon footprint falls, but there are other impacts that come with it.

119

You know, we won't talk about water, but water is a huge problem for data centers because they use it to cool and they are often built in places where there is water stress already.

120

So it's less about carbon from an AI perspective, but all the other stuff that matters.

121

What does this mean in the future, given that we're probably going to be relying on AI much, much more than we currently do?

122

Does that, even though the electricity is getting cleaner, like how do these two things balance out?

123

Well, as much as we are worried about carbon footprints, tech companies are worried about the cost of getting these searches served to you.

124

Right now, most people don't pay for the AI searches that they are doing.

125

And so they are really trying to cut their energy use down just to save money, if not to save carbon.

126

So we have already seen a lot of efficiency gains coming through in AI searches, some that Justin talked about, where they are familiarizing a model with the types of searches that have done in the past and then serve you a quick answer.

127

That will likely continue.

128

But so will the types of uses that AI can be put to.

129

Book your tickets now, maybe you'll start to write essays for school goers, all sorts of applications that we don't know.

130

So I think we are stuck in that place where people continue to use more and more AI and we are going to have to continue to find ways to cut its energy use and its carbon footprint.

131

Right.

132

OK.

133

So collectively, the impact is relatively small.

134

So individually, it's just absolutely tiny part of your carbon footprint.

135

Let's move on to our next question.

136

It's from Jim and he wrote to the climate question at BBC.com.

137

I listened to the latest episode and I smiled when Jordan, that's my co-host, said that climate change and earthquakes might go to the top of his nightmare list.

138

It made me think there is so much to worry about when we look at the state of our planet and hosting a show like yours must bring even more worries.

139

Jokes aside, I often feel that young people and all of us can feel overwhelmed about the future.

140

Now, Jim does go on to suggest we should make a programme about young people and climate, which we will do.

141

But I thought we could all just go around and give one reason for pessimism and one reason for optimism.

142

You can choose which way around you want to do it.

143

Justin, do you want to kick us off?

144

Let's start with pessimism.

145

I mean, when you're a climate journalist, as I am, your email inbox fills up literally on a daily basis with a silt of emails,

146

with really worrying kind of studies and reports from around the world about what's happening, not just in terms of the changing climate, but the impact of climate on ecosystems and other environmental destruction.

147

So, you know, you get a I mean, you know, a real kind of I mean, it's a proper doom scroll in the true sense of doom as you go through your inbox.

148

So you become acutely aware of the pressures there are on the world and the way that, you know, the climate's changing.

149

But on the other hand, equally, you know, there are some really positive developments.

150

I mean, we've touched on them already.

151

You know, potentially AI could be a hugely powerful force for making the world more efficient, making us use energy more efficiently,

152

finding new ways to use energy or create kind of, you know, the molecules that we need to, you know, fire factories and that kind of stuff.

153

So I think, you know, it's a real balance.

154

It's a balance between on the one hand, kind of, you know, pessimism and concern about the changes that are happening, how rapidly they're happening, you know, more than 30 years on.

155

And we're in the 31st years of the cycle of the UN climate talks.

156

And we're only, I mean, the good news is we're beginning to see, it seems, emissions begin to level off, concentrations still rising of CO2 in the atmosphere.

157

So, I mean, this gives you a sense of the mix.

158

We've reached a very high peak in terms of emissions.

159

If we do peak and start to reduce, we need to reduce really rapidly if we're going to avoid the worst impacts of climate change.

160

So it's very finely balanced, Graeme.

161

Mixed bag.

162

Caroline.

163

OK, optimistic.

164

A reason to be optimistic.

165

I guess in my job, I'm lucky in that I get to see and visit scientists who are doing amazing things to tackle climate change.

166

So, for example, earlier this year, I spent some time with scientists on the Great Barrier Reef in Australia who were growing heat resistant coral and planting them on the reef and already seeing tangible results and reducing coral bleaching.

167

So, yeah, those kind of examples make me feel hopeful.

168

And also, I think there is something in kind of choosing to be optimistic in a way.

169

I mean, research shows that there is a correlation between being generally optimistic and being physically and mentally healthier and happier.

170

And it kind of also feels like we kind of have to choose to be optimistic otherwise we won't do anything So even though I I would say my like my my core is sad Yeah, exactly.

171

I like to choose to be optimistic so I think that will help me make more of a difference There is actually a report on this and it's called the Goldilocks zone Mmm, okay where you've got enough sort of anxiety to motivate

172

But not too much so that you're paralyzed and not too little so you don't you kind of just you're like You make it sound like a very fragile pattern.

173

It feels like a fragile pattern.

174

So on the pessimism, I'd go one step further by saying, given climate change is the only problem for which there has been global governance of any sort.

175

You know, not one for pandemics, not one for AI robots, not one for nuclear weapons, really.

176

There isn't a global pact.

177

Some countries have a pact to try and reduce weapons.

178

Some countries just pulled out of that.

179

Exactly.

180

So if we can't solve climate change with all the facts that we know and a global governance system, well, I feel pessimistic about the other problems.

181

But on the optimism side, to me, climate has always been not just about climate change, but how much it can do good for the world, the types of solutions that we work on.

182

And one really tangible one right now, you know, we are living through what is the worst energy shock that the world has experienced.

183

This is the International Energy Agency telling us that.

184

and look at power prices in Europe.

185

They're actually pretty reasonable.

186

They've gone up.

187

It's not to say it's not higher than it used to be, but not as high as it was in the 2022 energy crisis.

188

And one of the key reasons is because at that time, we were able to install a lot of renewables and electrify a lot of our economies.

189

And that happened just four years ago.

190

So to me, the scale at which we can start to deploy many of these technologies and solutions have benefits outside of the climate problem.

191

And that's great.

192

Just on a pessimistic note on that, I mean, but it's still a tiny proportion of total energy use.

193

I mean, the scale of the transformation we need is absolutely huge.

194

But in a way that should, I think, exactly in the spirit of what you say, should be a source of optimism because we're in the middle of a really exciting industrial revolution that will create jobs and industry for millions, tens of millions, billions of people and a cleaner, you know, world.

195

So I think, you know, that's another strong optimistic point there.

196

Yeah, I think I get a lot of optimism from people who I meet on this, you know, through this line of work, doing really incredible, ingenious things every single day.

197

But anyway, we must round it off and finish there.

198

Thank you so much to Caroline Steele, Justin Rowlatt and Akshat Rati for joining us.

199

Thank you.

200

Cheers.

201

Thank you.

202

Thanks.

203

And thank you so much for watching.

204

Don't forget to subscribe to the BBC World Service YouTube channel so you don't miss future episodes.

205

and whilst you're there, drop us a comment, tell us what you like, don't like, perhaps you have a question for us and we'll feature it in a future show.

206

I'm Grae Jackson and I'll see you next time.

YouTube

4.9/5 on App Store & Google Play

Shadowing English
On Mobile

Learn English anytime, anywhere with the Shadowing English app. Improve your communication skills today!

Track your learning progress

AI grading and error correction

Rich video library

Scan To Download

Download on theApp Store

GET IT ONGoogle Play

What is the Shadowing Technique?

Shadowing is a science-backed language learning technique originally developed for professional interpreter training and popularized by polyglot Dr. Alexander Arguelles. The method is simple but powerful: you listen to native English audio and immediately repeat it out loud — like a shadow following the speaker with just a 1–2 second delay. Unlike passive listening or grammar drills, shadowing forces your brain and mouth muscles to simultaneously process and reproduce real speech patterns. Research shows it significantly improves pronunciation accuracy, intonation, rhythm, connected speech, listening comprehension, and speaking fluency — making it one of the most effective methods for IELTS Speaking preparation and real-world English communication.

☕ Buy us a coffee

ShadowingEnglish remains 100% free thanks to your support. Server and AI costs are high — your coffee keeps us going! 🙏

Donate via PayPal

ShadowingEnglish.com – English Shadowing Practice

Speak English fluently using the Shadowing technique. Listen to native YouTube videos, repeat sentence by sentence, and build real pronunciation and fluency — used by IELTS learners worldwide.

Shadowing Practice: How much energy do AI searches really use? - The Climate Question podcast, BBC World Service - Learn English Speaking with YouTube

Shadowing English
On Mobile

About This Lesson

Key Vocabulary & Phrases

Practice Tips

What is the Shadowing Technique?

Shadowing Practice: How much energy do AI searches really use? - The Climate Question podcast, BBC World Service - Learn English Speaking with YouTube

Shadowing English On Mobile

About This Lesson

Key Vocabulary & Phrases

Practice Tips

What is the Shadowing Technique?

Shadowing English
On Mobile