Are you hurting the planet every time you query a chatbot? Economist Meghan Busse suggests that may be the wrong question. In this episode, we zoom out from the usual concerns about frivolous queries (and, for that matter, resource‑hungry data centers) to ask an even more basic question: is intense demand for AI locking us into decades of additional fossil fuel use, or is it accelerating investment in clean energy? Busse helps us sort real climate levers from symbolic ones, and explains why AI’s ultimate impact on the planet will likely come down to how we build and power it.
**Further Reading: **
[Public attitudes toward climate policy, technology, and the environment](https://apnorc.org/projects/epic-climate-change-2025/?doing_wp_cron=1784231367.3495659828186035156250) — AP-NORC
[Utah Senate President Loses Primary After Data Center Backlash](https://www.nytimes.com/2026/06/24/us/j-stuart-adams-utah-senate-data-center.html) — Jack Healy, *New York Times*
Episode Transcript #
This is a computer-generated transcript. While our team has reviewed it, there may be errors.
Jess Love: This is Life, Automated, the show where we explore how technologies like AI are changing how we live, work, and make decisions.
This episode, we dive into one of the biggest anxieties around AI. What it’s doing to the environment.
[News footage from people protesting data centers…]
You see this anxiety most dramatically in the protests against data centers. It wasn’t that long ago that just a handful of people might show up at town halls to object to a project. These days, communities are getting more organized… and they’re increasingly calling for moratoriums on data center construction. That’s state and local freezes, basically. They’re citing worries about water, noise, pollution, land, and strain on the electrical grid.
[News footage from people protesting data centers…]
But I wanted to understand something more basic. Just how bad is AI for the planet? Particularly compared to all the other things we do to the planet?
And what should *I *actually do about it? Like me, personally, as an individual.
I have environmentally conscious friends who refuse to use tools like ChatGPT and Claude on principle, particularly for frivolous queries like generating a vacation packing list. I have a suspicion that this might be a little…beside the point? But maybe it isn’t?
So I reached out to Meghan Busse. She’s an economist at the Kellogg School of Management. She’s NOT an expert in data centers specifically, but she is an expert on energy, climate, environmental policy and how markets affect those things.
So I thought she’d be a great person to zoom out and really help me understand the big picture. How should I, someone who cares about the environment, be thinking about AI?
Meghan helped me separate the problem into two major buckets. The very real local impacts of a data center moving into your neighborhood, and the broader impacts on CO2 emissions that lead to climate change. And that second bucket is where the story gets really interesting. Because Meghan actually sees a future where demand for AI could benefit the planet. If we stay laser-focused on the right things.
I’m Jess Love, and this is Life, Automated, a show from Kellogg’s Ryan Institute on Complexity, distributed by KQED.
Meghan and I started our conversation discussing a recent AP NORC poll that suggested more than 70 percent of Americans are either somewhat, very, or extremely concerned about the environmental impact of AI. That’s more than the percentages concerned about the impact of air travel or meat production. I wanted to know what she made of this.
Meghan Busse: So on the one hand, I’m excited to have that many Americans concerned about the environmental and climate impact of anything. On the other hand, it is consistent with what I think we see in other places, which is that people sometimes have the biggest fears about the things that are either least familiar or the newest, or what I call as a technical term “scary.” And something that’s scary is something that makes a good disaster movie. It’s what makes nuclear power scary. But, coal-fired, air pollution not scary. Even though the coal-fired power pollution is much more dangerous in terms of mortality than nuclear power is. There’s the same kind of thing. I think that AI seems scary in the sense that we can imagine a great disaster movie about it when you don’t imagine a great disaster movie about the CO2 emissions that come from meat production.
**Jess Love: **I don’t think Meghan’s point here is that climate fears about AI are themselves irrational. But she’s already raising a question that will run through our whole conversation: are we worried about the right things, in the right proportions?
When it comes to AI, there are big global concerns around CO2 emissions related to energy use, both in terms of the energy required to train the models, and the energy required to run them. We’ll get to those.
But first I wanted to talk about THE thing people are most upset about. And that is what Meghan calls “local proximity issues”: the concerns we might have when we learn a data center is moving in nearby. Which in some ways are like lots of bad neighbor issues people have had with factories in the past….but with some new twists.
Meghan Busse: There’s a sense in which the pollution, if you will, that comes from data centers is different from what we have typically thought of when we think about industrial sources. And there we are oftentimes thinking about air pollution or water pollution or things that are contaminants that can harm people’s health because they get into people’s bodies, right?
Meghan Busse: And a data center is much cleaner in that sense. There isn’t smoke, right? There isn’t particulate matter that potentially people are breathing.
But people are worried, and I think with some justification, about other kinds of externality, other kinds of consequences. And one of the things that people have been increasingly talking about is the issue of noise. So you might be near a data center, and it wouldn’t be loud or clattery or things like that, but people describe sort of a hum or a buzz or a vibration-
Jess Love: Ugh
Meghan Busse: …Or a thrum, that can affect people’s health, not because it bothers them that they hear it, but because it is a low-level vibration that can affect people’s bodies in ways that I’m not sure we understand really well. And part of the challenge is that to the extent that we have noise ordinances… They are to deal with things that are loud, that affect hearing, that disturb quality of life in the way that we’re used to conventionally thinking about noise.
Jess Love: Right. It’s almost like we don’t even know enough about this to be able to design the regulations and the agreements that would then satisfy the community.
Meghan Busse: The more I read about this, the more I’m coming to believe that near people is just not a good place to put data centers.
Jess Love: So we talked about noise. What are some of the other concerns?
**Meghan Busse: **There are land use kinds of questions. So there is, for example, in northern Utah, in a place called Cache Valley near Logan, there was recently a proposal for an absolutely gigantic data center that would have been just a huge presence in the land there.
Jess Love [Voiceover]: I looked it up. The proposed hyperscale data center she is talking about would be located in Northwest Box Elder County, near Cache Valley. Its size: 40,000 acres. After a public backlash, the size of the proposed project area (that’s the area including both the facilities and the land) was reduced to 20,000 acres–a mere two-thirds the size of San Francisco. Having a neighbor that big…yeah, I can see how that could cause alarm.
And then there’s the question of water. Data centers have a reputation for being real water guzzlers, mostly because they so often rely on evaporative cooling to function. In places where water is scarce, they can compete with households, farms, and other businesses, driving prices way up…or leading to shortages. And while there are some ways to make water use more efficient, these come with their own tradeoffs, like requiring a lot more energy.
Still, as challenging as these “local proximity issues” are to deal with–Meghan says they are NOT what concerns her most about AI’s potential impact on the climate. In part because they SHOULD be solvable.
Meghan Busse: I think that many of the proximity concerns could be solved by altering where data centers are built. There’s a lot of parts of this country that don’t have a lot of people in them, right? There are less dense, less populated places that could be chosen for data centers.
Jess Love: But why aren’t they?
Meghan Busse: Well, it’s not as convenient.
Jess Love: It’ll cost more money.
Meghan Busse: I don’t know if the land will cost more money but it’s not as easily connected. It’s not necessarily where the eventual employees that you wanna have running the data center or interacting with the data center are most easily located.
Jess Love [Voiceover]: So I can see why people are so angry about having a data center as a next-door neighbor.
But what makes data centers bad neighbors is not necessarily what makes them so bad for the climate overall. What Meghan Busse–a researcher who studies climate policy–is most concerned about … is energy.
Meghan Busse: So I think that the electricity use might be the thing that is most distinctive about AI. And the reason for this is that electricity use in the US and the EU and these sort of high income developed countries, was increasing pretty steadily along with population and GDP from post World War II until about 2000. And then it flattened out. And this isn’t because people didn’t want to use electricity for things. It’s because the devices that they used for electricity became more energy efficient. So we could get more of what economists call energy services without needing as much electricity. But what that meant is that, for about 20 years, there wasn’t a lot of need to greatly expand or to really expand very much at all.
Electricity transmission, electricity distribution, the total amount of electricity generation that we could do. And then all of a sudden along comes AI which is really pretty energy intensive… electricity demand, and it’s coming at the same time that there’s an increase in vehicle electrification and an increase of automation and manufacturing. And it’s putting a real strain on the ability of electricity systems to meet that.
Jess Love [Voiceover]: So. Here in the U.S., we have a couple decades where the overall amount of electricity we need to generate basically stays flat. Electricity goes into a lot more THINGS, but all those THINGS get more efficient at about the same rate. That’s two decades of not a lot of pressure to build new power plants. And then…BAM.
And the timing of this new strain on the electrical grid is relevant because there’s been another trend at play.
**Meghan Busse: **Over the last decade or 15 years, the cost of generating electricity with renewable sources, with wind and solar has fallen dramatically. So over the course of about 10 years, the cost of generating electricity with solar went down to about a tenth of what it was, and the cost of generating electricity with wind had gone down to about a third of what it was.
Jess Love: That’s a tenth. That means if it used to cost a dollar, it costs 10 cents.
Meghan Busse: That is right. It’s a phenomenal reduction. It’s a phenomenal achievement for the climate. It might be the single best event to happen for the climate in the last 10 or 15 years. It’s just phenomenal. And it is so phenomenal that what was happening is that basically electricity generators all over the country and also in many places in the world, were in a process of transition, in a process of retiring coal-fired power plants and replacing them with solar and wind – not because they were greener, but because they were actually cheaper.
Jess Love [Voiceover]: Thanks to just how much the economics here have changed, nearly all new electricity generators planned for the future in North America were renewables, with a bit of natural gas, but basically NO coal. This was going to be a huge win for the planet. Even better, coal plants were actively being PHASED OUT. If you cared about CO2 emissions, this was very good news. Everything was headed in the right direction.
Then, out of nowhere, for the first time in forever, demand skyrockets. All those plans…poof!
Because these data centers – they don’t just need a ton of energy. They need it fast.
If you’re an AI company, you want to get your new data centers up and running as quickly as possible. You feel like you’re in a race: to create the best models, to capture the market. Sure, in the long run, over 20 or 25 years, solar and wind are cheaper. But if you need power NOW, keeping an aging coal plant running a few more years is going to be pretty tempting. And that smooth march toward cleaner power just got a lot choppier.
Coming up: the SHORT-run danger to the climate is that AI keeps fossil fuels around longer. But in the long run? Well, the jury is still out. Economist Meghan Busse offers up a scenario where the AI data center boom could actually benefit the planet. That’s after the break.
[ad break] **Jess Love: **So I wanna turn now to the future. I think you’re completely right that there is this sense that, you know, we’re straining under today’s demand and everybody is telling us AI is going to be so many times bigger in a couple of years. What the heck happens? How are we gonna keep up? And how does this not like, destroy the planet?
**Meghan Busse: **So if we take the long run view of the climate, many climate scholars would tell you a key component of decarbonization is to electrify as much as we can. And the reason is because we have really good ways of producing electricity without producing carbon. So solar and wind are two of them that we’ve talked about.
But nuclear energy also has the potential to produce a lot of electricity in a decarbonized way. And there’s other possibilities also out there. Things like geothermal that, you know, people are experimenting with and hoping will come up to scale. But, there are, you know, like I say, wind and solar that we already have, nuclear that we also have already, although we need to re-remember how to build it at low cost.** **
Jess** Love:** Just jumping in here to say, if you’re over 40, and you care about the environment, you might remember a time when environmentalists were dead set against nuclear power. There was that deadly catastrophic meltdown at Chernobyl in 1986, and a scary miss at Three Mile Island a few years earlier. Environmentalists were also concerned about the huge challenge of storing toxic nuclear waste for thousands of years.
But in the last forty years, many scientists agree that nuclear power and storage have gotten much safer and less unwieldy. And some environmentalists now feel that the benefits of nuclear power–like not spewing carbon into the atmosphere–well outweigh the risks.
Meghan Busse: And there’s other kinds of technologies, like carbon capture that might allow us to burn fossil fuels and still capture the carbon and have it be decarbonized. So we’ve got some really great potential for being able to produce decarbonized electricity.
And so what that means is that in the long run, I’m not that concerned from a climate perspective about anything that uses electricity. So if there were a new technology to come about that needs a lot more energy and you told me “guess what the energy it needs is electricity”, then I kind of heave a sigh of relief as opposed to saying what we’re gonna need in the future is a lot more steel or what we’re gonna need in the future is a lot more cement. Or what we’re gonna need in the future is a lot more air travel because those are things we don’t yet know how to decarbonize very well or at very low cost.
Jess Love: Okay. So in the long run, because it can be electrified and because we know that there are renewable ways of getting electricity, less concern. How long is the long run? And I guess how much damage is done along the way?
Meghan Busse: It’s a really great question because some of the answer to how long is the long run depends on how we respond in the short run. If the way we respond to the intense increase in demand is the short run, is building new fossil fuel electricity generation, like building new gas plants, those plants have a long life, 20 or 30 years. Things that exist tend to get used. And so if we respond to the short run pressure by making investments in new fossil fuel generation, then that’s gonna push that long run of decarbonization off farther, and so, for the climate, the best news scenario in some ways would be if builders of AI and data centers decided, well, wait a minute, the most reliable and fastest way for us to have the electricity we need is in some sense to be self-sufficient in a green way, to build a data center, to build accompanying, solar or wind generation, to build a battery facility that will help make sure that that power is available to us throughout the day.
Jess Love: Well, and some of this is happening, so I just read, Google is making plans to purchase Intersect, which is a wind and solar company bringing energy generation in house and almost kind of treating it as a competitive advantage.
Meghan Busse: Absolutely. And Microsoft recently made an agreement to reopen one of the reactors at Three Mile Island. Again to do essentially self-supply with nuclear energy. So to the extent that the response to this is self-supply with some kind of green energy, that is number one, good for the climate, but potentially also good for, accelerating storage technologies, for accelerating our relearning how to build nuclear in a competitive way. Those would be the best case scenarios.
Jess Love: Accelerating because it’s really kind of like, forcing people to innovate and act very quickly to do the kinds of things that they might otherwise do gradually over the course of 10, 20 years.
**Meghan Busse: **And what drives down costs in many, many technologies is, increasing the scale of demand, basically increasing the size of the market. When you go from, here’s a technology we have thought of to one we have tested in the lab and demonstrated to one that we have done pilot projects of and works, there is this important phase between that and sort of a large scale deployment, which is increasing the scale of production enough that costs fall. And I totally believe it’s something we’ll get better at as we do it more.
Jess Love: Yeah, and I think that’s a really, really interesting point. Like there is a world in which we can harness all of this demand and all of this excitement for AI and actually leave the climate in a better place than it would’ve otherwise been.
Meghan Busse: And I, you know, and I’m not saying this is guaranteed, I’m not saying that this is obviously what’s gonna happen, but it like, there is a path and it actually has the potential to accelerate you know, green energy and decarbonisation in ways it would be really positive for the climate.
[musical interlude] ** Jess Love [Voiceover]: **So, Meghan does see a future where demand for AI helps to accelerate deployment of low-cost green energy. But that’s not the only optimistic claim people make about AI and the climate.
**Jess Love: **Okay. So I wanna do a kind of nerdy lightning round with you. So I’m gonna share some other ways that AI might impact the climate in a positive way. And, we’ll say zero is like, that is all just hand-wavy gobbledygook, and five is like, yeah, actually there could really be something there. Okay, you ready?
**Meghan Busse: **Ready.
Jess Love: ‘Optimization of the energy grid.’
Meghan Busse: I’ll give that a four.
Jess Love: It’s pretty good.
Meghan Busse: I’ll give it a four because it is the sort of thing that AI can do. Right. There’s already predictive maintenance being done and AI being applied to in a lot of different, manufacturing and other kinds of things. That said, it’s actually not the most important thing, I think, in the optimization of our energy system. We just need way more transmission in way more places to accommodate renewables generated in places where we historically have not had electricity generation like the Southwest. And just greater supply and greater flexibility. So I’ll give it a four ’cause AI can do it, but it’s actually not the most important thing you could do if you wanted to improve the electricity transmission grid.
Jess Love: That was very thorough.
**Meghan Busse: **Not very lightning, but very thorough.
Jess Love: Very thorough. ‘Better climate modeling.’
Meghan Busse: There’s a sense in which I don’t know whether you really want AI climate models. In the sense that part of the trick about AI is that it’s a bit of a black box. Climate models are basically systems of equations. Systems of equations that capture relationships we know about heat and temperature and weather and ocean currents, and then how those translate into economic consequences. Part of the value of climate models is that we know what assumptions we have put in and we can alter those assumptions in ways that allow us to see if this were different, how would that matter for our climate predictions? I’m not sure that sort of turning that over to the more black box nature of AI is gonna improve our confidence. So I’ll give that a one.
Jess Love: ‘Discovery of more energy efficient materials.’
Meghan Busse: That’s a great question. So there are absolutely applications in which materials matter – Battery chemistry, for example, is one that a lot of people are pushing on. So I’ll give that a three.
Jess Love: ‘ Precision agriculture’, which is this idea that with the right sensors and models, we could give crops exactly what they need when they need it, improving yield, reducing waste.
Meghan Busse: So I’m gonna give that a one, not because I don’t think AI can do it, but because, doing it more efficiently or more precisely doesn’t do a lot for decarbonization if we’re using the same techniques, in terms of fertilizer, in terms of planting, in terms of those kinds of things. So I think in order to make a big difference in agricultural emissions, we’re gonna need something other than, sort of more precisely, implementing the technologies that we have.
**Jess Love [Voiceover]: **Ok, so Meghan doesn’t see a lot of potential in some of these super optimistic “AI will just create new things that help!” scenarios. But she does see the potential for AI to fuel a clean-energy tech boom. At least…in the future. But how can people like me make a difference TODAY? SHOULD I feel guilty asking ChatGPT for that packing list?
Meghan Busse: So, one of the things, as I teach classes about climate, and as I talk to audiences about climate, I oftentimes get questions about, what can I do that matters?
And I don’t think people always have a very good idea of what are the first order things, like what are the big levers they have in their hand, and what are the small levers that they have in their hand? If you ask me, I would tell you that generating a packing list on ChatGPT for your vacation is not a very big climate lever. In fact, it’s a much bigger climate lever how far away you are going on your vacation and how are you going to get there? If you’re taking an airplane, the climate impact of your airplane trip is many, many, many, many, many ChatGPT queries. So I think part of the reason people have fixated on ChatGPT queries as something that is part of their climate virtual signaling, is that it seems like an easy thing to do or not do as opposed to deciding not to drive to work or not to have a furnace in my house, or…
Jess Love: – not to go on vacation.
Meghan Busse: Not to go on vacation. And so we sometimes get fixated on the things that seem like, well, that just seems like an excessive choice that you are making. But if you wanna know the things that really matter, the things that really matter have to do with your more direct use of fossil fuels. Think about how you set your thermostat. Think about what you drive and where you drive. Think about where you choose to fly. Think about what you eat – Beef is especially carbon intensive. So if people are asking me, what is a marginal choice I could make in my life that would make a difference, using ChatGPT would not be at the top of my list.
**Jess Love [Voiceover]: **Well, what about…joining some sort of collective action effort–like the protests we’re seeing against data centers? That strategy seems to have more potential, at least for those local issues. Because all data centers are not created equal, and being able to shape the kind of neighbor you have, and where it situates itself – that feels big.
Jess Love: Are communities able to get some concessions from these companies that allow the data centers to sort of be built, you know, somewhere in the proximity, but in ways that don’t impact people’s quality of life as much?
Meghan Busse: So I think there’s actually a ton of heterogeneity as you go from place to place. I mean, it’s very different if you have something that is federally regulated, something like, you know, pharmaceuticals or those kinds of things, right? The process might be slow, but once a decision gets made for something that’s federally regulated, then that’s the decision, right?
And when you have something that, that it’s gonna be sort of a location and state by state, it’s just gonna be a lot noisier, it’s gonna be a lot less predictable, and I think that, what’s gonna turn out to be a successful strategy for communities to employ is gonna depend a lot on the particular location where they are.
**Jess Love: **And perhaps what they want to get out of it.
Meghan Busse: And perhaps what they wanna get out of it, which may not be the same place to place.
Jess Love: Yeah, I mean, could this be a good thing, where communities, you know, depending on the tools they have available to them, can sort of look at the situation and in a bespoke way try to come to an agreement that is going to be as much of a win-win as possible?
**Meghan Busse: **I hope so.
[music bump] Jess Love [Voiceover]: It’s clear that local opposition to data centers can have an impact. Meghan sent me a newspaper article involving a powerful Utah State Senator. He served as chairman of the agency that approved that giant data center we talked about earlier, the one that was downsized to a mere 20,000 acres. And…he recently lost his Republican primary bid, in what the article called, “one of the most high-profile signs of the voter backlash to data center projects.”
Which brings us to the final thing that individuals can do–something that works for these local issues and for the climate more broadly.
Meghan Busse: If you care about having an individual impact, by far the most impactful thing you can do is how you vote.
Jess Love: There you go.
Meghan Busse: And that’s because so many of the changes you need are not about our individual consumption or our individual behavior. It’s about the systems of how we generate energy. The systems of how we use energy, what we use energy for. The things that are really, really hard for individuals to influence with their choices.
So in some sense, I don’t care very much about whether you use ChatGPT, I don’t care very much about whether you eat beef or not. I don’t even care about where you fly on vacation as long as you vote.
Jess Love: So you teach the next generation of CEOs, C-Suite Executives, and you have children of your own. What advice do you have for young people right now?
Meghan Busse: The advice that I have for young people now is that while it seems, like on the political front, it might seem like things are discouraging on the technology and economic front, there’s huge potential.
We can absolutely decarbonize, we can absolutely avoid the worst effects of climate change. We have many of the technologies and I totally believe that we can invent the technologies that we still need. It is absolutely within our power if we choose to do it.
**Jess Love [Voiceover]: **So. In the long run, how will AI impact the climate? It depends. But after my conversation with Meghan, I’m thinking: it doesn’t depend that much on the AI itself.
It depends more on what we do with demand for the technology. Where and how companies build out their data centers and especially how they choose to power them.
Are we going to live in a world where demand pushes back our transition to clean energy, locking us into fossil fuels for decades? We’re seeing some of that world right now.
But we can also choose a world where this explosion of demand hastens our transition to clean energy. And we’re also seeing some of that right now.
The politics of clean energy can look grim…But as long as clean-energy technologies keep getting cheaper and easier to scale, the optimistic scenario is still on the table.
This is Life Automated, a project of the Ryan Institute on Complexity at the Kellogg School of Management at Northwestern University. Distributed by KQED. I’m Jess Love. Special thanks to today’s guest, Meghan Busse, an economist and associate professor at the Kellogg School of Management.
We’re produced by Jesse Dukes. Music by Steven Jackson. Recording by George Christensen and Will Feeney. Support and wise counsel from Stacia Sliger.
END