I ask them and they essentially have the same discourse saying, doesn't matter, we're working on AGI. And for me, as like a scientist, I find that really shocking because it's like, well, yeah, you have scientific responsibility. You should be transparent about the cost of your work. That's part of being a scientist. That's part of being a researcher. And the fact that it's percolated to that level, yeah, really worries me.
SPEAKER_00Sasha is an artificial intelligence researcher and climate lead at Hugging Face. Now, you've probably seen plenty of stories in the past year or so about the rising emissions at companies like Microsoft and Google, the large water use of their data center networks, and of course the growing concerns about the energy use and computational demands of generative AI tools released by OpenAI, Google, Meta, and these other major tech companies. At the same time as temperatures seem to be rising, as the natural disasters that we experience globally seem to be getting worse because of climate change, the tech industry has decided to massively increase its energy use and its computational demand to build out many new hyperscale data centers in order to power these AI ambitions with little thought for the climate impacts and the sustainability goals that they're breaching by chasing their business interests, by trying to grow their cloud computing businesses, and by rolling out generative AI tools that don't actually provide the benefits that they claim they do. The climate is being further ravaged, and our ability to actually stay within the targets that we need to meet are being made more and more difficult because the tech industry, not just these major companies, are arguing increasingly that we need to be massively increasing our energy use, or else we as a society are going to stagnate and life will become a lot worse. But as we discussed with Jason Hickel a few weeks ago, that is not necessarily the case. And there are other alternatives that we can pursue in order to build a better world. But that is not in the interest of companies like Microsoft and Google. So, with that said, Sasha is on the show today to talk about this question of the energy use and the computational demands of generative AI tools, why it is so high, and why it doesn't make sense often to roll these tools out in the types of use cases where these major companies are trying to embed them right now. This is really a wide-ranging conversation, but also a really fun conversation. I've known Sasha for a little while, and I think you'll probably hear that in the conversation where we have a really good back and forth on some of these issues. It helps, of course, that not only is she very knowledgeable about these things, but this is an area where I have been spending a lot of time reading a lot more and learning a lot more about what has been going on here. So it's easy for me to kind of pull different examples and pull from things that I know. So yeah, that's just to say that this was a lot of fun. I think that you're going to enjoy it as well. And I think that hopefully you'll learn quite a bit from the discussion that we have too, because this is obviously a very important issue. And I was debating whether to do an episode on it specifically where, you know, we are planning to do a data center series in the fall, but I decided it's important enough that, you know, we should just do this now. We should have this discussion. And then if we need to pick up threads again in the fall, that's totally okay to do. So if you do enjoy this conversation, make sure to leave a five-star review on the podcast platform of your choice. You can also share the show on social media or with any friends or colleagues who you think would learn from it. And if you want to make sure we can keep having these critical, in-depth conversations that we're not replaced by chat bots and that it's actually humans here chatting about these issues, you can join supporters like Andrew from Queens, Oscar from Australia, and Ryan from St. Louis, Missouri by going to patreon.com slash TechWon't Save Us, where you can become a supporter as well. Thanks so much and enjoy this week's conversation. Sasha, welcome to Tech Won't Save Us.
SPEAKER_01Thanks for having me.
SPEAKER_00Absolutely. Very excited to chat with you. I know that you've been researching this topic for a long time, but it's one that is very important in this moment. We've been talking about having this conversation for a while. But recently there were these figures that Microsoft and Google revealed that, you know, despite making net zero and carbon neutrality pledges back in 2019, their emissions are soaring, right? Microsoft's are up 30% between 2020 and 2023, while Google's are up 48% in five years. And they say that a lot of that is related to, you know, this kind of AI rush that they have been in for a while now. What was your reaction when you heard those numbers from those major companies?
SPEAKER_01Honestly, I was expecting something like that. I was surprised by the fact that they were actually transparent about it, because usually this kinds of stuff gets like swept under the rug or put in the footnotes or something. But in this case, it was pretty straightforward. And I mean, it's not that surprising if you're following the evolution like I've been for the last like what, seven years now? It's like it's growing, the environmental impacts are growing. But the thing is, is that like with generative AI, it really kind of kicked things into overdrive. And so when these promises, when these goals were made, generative AI wasn't really a thing. And so it's like, of course, this is really like blew it out of the park because nowadays, like Gen AI is being put into everything and anything. And so, of course, it comes with energy costs. But also, you know, unsurprisingly, like most countries and and companies don't really meet their climate targets. Like, I mean, it's there's a lot of fanfare when they're put forward. There's a lot of kind of greenwashing or signaling. And then when it comes to the actual date, it's like, oh, well, we overshot it, or oh, well, things changed, or we didn't consider this, or we're gonna change the way we calculate this. And so this is actually part of like normal ESG reporting.
SPEAKER_00Yeah, and what you say is exactly what Microsoft said afterward, because when they announced their big like carbon neutrality pledge, they called it a carbon moonshot. They were gonna hit this big goal by 2030 of having negative carbon emissions. And then after this news came out like last week or whenever it was, Brad Smith, the president of Microsoft, said, quote, the moon is five times as far away as it was in 2020. If you just think of our own forecast for the expansion of AI and its electrical needs, it's like, okay, this moonshot we were planning is like really a lot more difficult to achieve because our business goals have changed and now we need to like race and try to capture this like AI market and the cloud opportunities that it offers as a result.
SPEAKER_01Yeah, and also companies like Microsoft are also not only doing in-house AI and generative AI, but they're also providing compute not only to open AI, but to customers, like above and beyond that. So I think that they're in a particular pickle because it's like, well, they're stuffing it into Word or Office or whatnot, and they have to provide the compute to open AI and they're training. And also customers want to jump onto the bandwagon. So it's like, I think that they're particularly in a corner and yeah, five the moon being five times further away is really accurate.
SPEAKER_00Yeah. And when you say in a corner, they mean like, you mean like with how they have to report their emissions and like what they're doing on the climate front. Business-wise, they're like, this is great for us. We love it, you know, we're expanding our businesses.
SPEAKER_01Well, I mean, they're in a corner because I think that shit will hit the fan at some point because they need to build more data centers. They need to have like more uh power purchase agreements if they want to at least keep some semblance of so power purchase agreements is essentially to buy renewable energy with like actual specific providers. Like I have a solar farm and I can sign a power purchase agreement with a data center kind of stuff. And so I think that they like they have a lot more targets, like they have a lot more things they have to do now in order to keep this machine turning, because it's like it's not only up to them, right? It's also up to the customers. And because, of course, they're doing so well, they have to like they can't all of a sudden be like, no, we can't give you guys more NGPUs because we're out of data centers, right? That would be pretty catastrophic. So it's like this this self-fulfilling prophecy that that's kind of putting them in a hot spot.
SPEAKER_00You have to wonder as well, like, you know, obviously we've been talking about this AI moment as like a bubble or a boom, knowing that there's going to be a moment when things kind of come back down to earth. And you have to wonder like what that does to these ambitions that a company like Microsoft has once that period kind of shifts and we head back down toward like a more normal like discussion of AI or like another AI winter or something like that, rather than like all the hype and the excitement that we've had for the past year and a half or so.
SPEAKER_01I'm just worried that once all the dust settles, if the dust settles, if there's no like new paradigm that gets invented in the meantime, that you know, we're gonna look back and be like, oh, oops, like that was a lot more like carbon than we expected. And I mean, historically as a species, we have a tendency to do that like retroactively, look back and be like, oh, this was worse than for the planet than we expected, right? And I'm afraid with AI, it's that because when I have these conversations with people who work at these companies, they're like, well, you know, first of all, for example, we don't really know what people are running on our data centers. And so we just provide the compute and they could be doing whatever, and we don't really like necessarily track energy use and like just a high level, yeah, but not process-wise and things like that. And so when I'm like, give me some numbers, they're like, we don't have the numbers. And I'm worried that by the time they get the numbers, we're gonna be like, whoa, this is even worse than we thought.
SPEAKER_00Yeah, and you know, this moment with AI, I feel like it really makes you think of the discussions that we were having a few years ago about Bitcoin and cryptocurrency and how much energy and how much computation they were requiring for their proof of work, like methods of kind of processing these transactions and whatnot. And how, you know, there was a lot of data center power, there was a lot of computation, there were a lot of GPUs that were needed in order to power all those things. And it feels like we've moved from recognizing that, okay, this was a real problem. There were a lot of emissions that were associated with that that we didn't feel was like worth it for what these technologies were really providing. And now we've entered this stage where like we have these generative AI tools that these companies want to be everywhere. You know, there's a different kind of aspect to it where with the cryptocurrency stuff, it was generally like newer companies, companies that were starting up, you know, they were not like the established firms that we have today. Whereas with generative AI, sure, there's the open AIs and there's the anthropics and things like that. But like Google and Microsoft and Amazon are like key in that whole ecosystem and like not just driving it, but benefiting from it when you think of their data centers and stuff as well, right?
SPEAKER_01Yeah, and I think that also as opposed to Bitcoin, like Bitcoin is relatively contained. Like I want to say that at least like you can figure out, um, for example, like usually it's it's concentrated or like you have some idea of how much energy it's using, but AI is like is literally, you know, they have the internet of things, you have cell phones, you have all sorts of like you have on device, you have on cloud, you have so much, like so many different components that I don't think we're getting the whole picture. And also companies that use AI, like when you talk to just like your average small or medium enterprise, they're like, we have a chatbot now, or we have this, we have that, they don't necessarily like take those numbers into account when they're doing their like carbon accounting or energy accounting. And when you talk to them about it and they're like, oh yeah, we switched our you know, good old-fashioned AI search system to generative AI, and I'm like, okay, well, is there like some numbers that you're getting? And they're like, no, like we're just paying our AWS bills or our Azure bills. And then when you we really tell them that you know this is not just like free compute, ephemeral compute, they're legitimately surprised because they also have these ASG goals and they're not actually accounting for any of this stuff when it comes to their ESG goals, even in the future.
SPEAKER_00Yeah, it's worrying, but it's not surprising, right? I wanted to kind of shift our conversation a little bit to understand a bit more about how this generative AI moment is really driving like energy use and growing emissions. So, how do these generative AI models actually use so much energy? Like, why is that?
SPEAKER_01So if you think about it fundamentally speaking, if you compare a system that uses, I guess, extractive AI or good old-fashioned AI in order to get you answers, for example, to search the internet and to find you an answer to your question, it's essentially what it's doing is that it's converting all these documents, all these like web pages from words to numbers and it's like vectorizing them essentially. And when you're searching for a query, like, I don't know, what's the capital of Canada? It will also convert that query into numbers using the same system. And then matching numbers is like super efficient. So if you're trying to find essentially like similarity, cosine similarity, like it's just like this stuff goes really, really fast. Actually, my very first job, I was helping like develop these kinds of systems, and it's like it uses no compute at all. It's like it can run on your laptop, it can run anywhere. But if you're using generative AI for that same task, so like finding what the capital of Canada is, instead of finding existing text numbers, it's actually generating the text from scratch. And I guess the advantage, quote unquote, is that instead of just getting Ottawa, you'll get like maybe a full sentence, like the capital of Canada is Ottawa. But on the flip side, it's actually generating like the model, the AI model is generating each one of these words sequentially. And so like each bit of that uses compute. And so the longer the sentence, the output, the more compute it uses. And you know, when you think about it, for tasks, especially like question answering, like finding information on the internet, you don't need to make stuff up from scratch. You don't need to generate things, you need to extract things, right? So I think fundamentally speaking, what bothers me is that like we're switching from extractive to generative AI for tasks that are not meant for that, right? Like just fundamentally speaking. And of course, there's hallucinations and like how many rocks a day do you need to eat and whatnot, but you know, also the high energy costs. So for me, none of this makes sense.
SPEAKER_00Exactly. As you were describing, like the generating versus the just like coming up with what's there. That's exactly what I was thinking of too, right? Like I, when I'm trying to figure out, you know, say the capital of Canada or something, I don't want the search engine to start talking about, you know, the glue that is going to go in my pizza or something like that, right? Like just give me the capital, just like respond to my query with something that's already there, you know, with these search results. Why do you need to start like making stuff up and writing these paragraphs just because it's like in vogue right now and the thing that a lot of these companies are pushing as like the next big thing when it doesn't provide like the utility that we're actually looking for?
SPEAKER_01Yeah. And like nowadays, when you search like Google, you have these like AI summaries, and a lot of the time they're false, but also like they won't show you like the actual references. Like me often, I mean, maybe for the Campbell of Canada is more basic, but often if I have like a more specific question, like what is the energy use of data centers worldwide? I want numbers, I want like citations, I want the actual like I want the receipts, right? And then if there's like this little resume, first of all, you I mean, I particularly don't trust them. So like they could be putting thing is generative AI is essentially based on probabilities. And so it's gonna generate text that's very plausible, that like sounds really well, but it's not necessarily like true in itself. There's no concept of like underlying truth to all this. And so on the one hand, you do have this high energy cost, but on the other hand, just like information-wise, it's not, it doesn't make sense for a web search particularly.
SPEAKER_00Yeah, it and that makes a lot of sense, right? Because it's not what you're really looking for when, especially you're trying to do a web search or something like that. I'm wondering as well, you know, we talk a lot about the models, but also the use cases for these AI tools, right? On the one hand, you need to train these big models that are going to actually do the work of generating this text or generating images or whatever. And then you have the actual use case where you go to Chat GPT and ask it to generate something, or you know, you have Google churn out one of these AI overviews, or you go to DALI and have it churn out an image. What is the difference there between training the model itself and then the actual using of the tool once that model has been trained?
SPEAKER_01So it's interesting because initially most of the studies about energy usage and carbon emissions were really based on training because everyone's like, well, so training typically, for example, for a large language model that's like billions of parameters, it's parallelized, right? So it's going to be like a thousand GPUs for something like a month or three months sometimes. So it really does add up. Like it can be like a year in total of compute if you actually like used one GPU. And so people were really focusing on training because it seemed like the elephant in the room. But then nowadays, like less and less people can actually afford to train these models, these large language models, but more and more people want to use them. And so now we're thinking about deployment more and more. And so in the recent study that I led, we were really comparing like the energy consumption of training versus inference. And depending on the size of the model, it would, it was between like 200 to 500 million queries, which seems like a lot. But if you think about it, Chat GPT gets like tens of millions of queries per day. So within like a couple of weeks, depending on the size of the underlying model, you'll have used as much energy as training the thing.
SPEAKER_00Yeah. And so I guess initially, especially when these tools were more novel and you know, not as many people were using them, the focus was more on what was going into actually training this large model, this general model that was doing a lot of things. But now that it's actually out into the world and millions of people, if not billions, are starting to interact with these things and use them regularly, and they're being built into so much of like the infrastructure of, say, the platforms that we use when we go online, now the actual using of these tools is the thing that is providing like the worrying compute demands and energy usage and all these sorts of things, right?
SPEAKER_01100%. And also what I really worry about is really kind of like the delta. So, for example, when you're switching between, I don't know, like a good old-fashioned extractive AI model to a generative one, like how many times more energy are you using? And in that study, we found that, for example, for question answering, there's like 30 times more energy for the same task for like answering a question. And so what I really think about is like the fact that so many tools are being switched out to generative AI, like what kind of cost does that have? And we don't really see those numbers. Like someone recently was like, oh, I don't even use my calculator anymore. I just use Chad GPT. And I'm like, well, that's probably like 50,000 times more energy. Like I don't have the actual number, but you know, like a solar-powered calculator versus like this huge large language model. So that's what like keeps me up at night is this really like this switch in all these different uh like nowadays people are like, I'm not even gonna search the web, I'm gonna ask ChatGPT, I'm not gonna use a calculator, right? All of that, what the cost to the planet is.
SPEAKER_00Yeah, that that's wild, especially when you think like these large language models in ChatGPT are not really designed to do math either. It's like, is your math even gonna be right?
SPEAKER_01I know, I know, I know. And and and but it's the thing, it's like they're marketed as these like general purpose technologies. There's actually this very confusing paper that's called like GPTs are GPTs. And so GPT, like in the open AI sense, it stands for generative pre-trained transformer. And GPT in that paper also is stands for general purpose technology. Anyway, it's a space paper that uses those two acronyms interchangeably to say that like the transformers are general purpose technologies that can do anything. And the point is just like essentially like you can use them for answering questions and making recipes and you know, math and whatever, and answering your homework, for example, right? And then and it's really like there's this push to say that they can do anything without any kind of transparency with regards to the cost.
SPEAKER_00Yeah, that that's so wild. You know, when you think about the energy cost of these things and what actually goes into making them, all the data that they're also trained on, right? Because we know that for a lot of these companies, it depends on having these vast stores of data that they've taken off of the internet in order to train these large, as you say, kind of general models that where the assumption is they can do virtually anything. Is that a consideration in this as well?
SPEAKER_01So there's really little study that has been done on like the trade-off between the amount of data you have and you know how much compute it uses. I mean, there's like these general kind of, they call them scaling laws, and essentially they've been driving uh especially like the language side of AI, but actually images too. It's like the more data you have, the better your model will be. And then there's been a couple of studies that showed this for like specific use cases. And so we've been pursuing this paradigm, like the bigger is better paradigm for like five years now, essentially. And everyone, like no one really even questions it. And I feel like it's like bigger is better in terms of data. Like the more data you can get your hands on, the better your model will be. But also for model size, like people literally would be like, let's add a billion or 10 billion parameters just because like it's definitely gonna be better. It's definitely gonna beat the benchmarks. So there's like this general kind of like rat race with regards to size in machine learning.
SPEAKER_00Yeah, and I feel like that plays into the conversation that people have been having over the past year or so. When you know, OpenAI releases new versions of ChatGPT and people are like, is this even better? I feel like it doesn't even work as good as it did before. Like there's a really kind of calling into question whether this idea of bigger is better really makes sense and delivers, you know, real benefits in the long run that most people can see or notice, or even if it happens at all, right?
SPEAKER_01And how we evaluate these models is so broken. Like most of the time, like these benchmarks, right? Like they're most likely based off of data that's probably in the training set somewhere. And so, for example, it's like, oh, there's this benchmark that's like, I don't know, whatever the bar exam, the whatever, the New York State bar exam. And then it was like, oh, Chat GPT pass the bar. But like the chances of that data being somewhere in GPT, whatever's uh training corpus are really high because probably there's some like books and study PDFs and whatever on the internet, right? And then there's absolutely no due diligence that's being done. It's just like you test your model and it passes all these benchmarks and you're like, look, it's awesome. It's it reaches human performance. And then like there's no kind of cause-effect situation. Like you don't try to figure out why that is. Is it because it actually like learned some useful patterns or like some logic, or because it just memorized the training data?
SPEAKER_00Yeah. Yeah. I I remember being so frustrated by those stories because I was like, it's not even just like an open book exam. It's like all the answers are like written next to the exam for you to copy over. Like, but that kind of bleeds into the misconceptions that can exist around these things based on. The way that the companies talk about them and promote them, and the way that you know, then the media kind of repeats those narratives and the public, you know, ingests it, right? Because they don't have a technical understanding of what is actually going on here.
SPEAKER_01Exactly. And where the emphasis is as well, like on the one hand, you've got like, this is gonna solve everything, this is gonna do everything. And on the other hand, it's like, well, this is potentially gonna wipe out humanity. But then none of that conversation is ever around like, what are the labor costs? What are the environmental costs? Like, like, for example, when ChatGPT really came out, and like I read the report or however they call it, what struck me is that like the paradigm didn't shift, like the actual like large language model paradigm transformers was just the same. But the amount of like human labor, like the amount of crowdsourcing that went into that was just like some something that nobody ever did before. And of course, it's hard to say whether that was like the magic ingredient, the secret sauce, but you know, that was really for me, like from a technical perspective, that was the difference with regards to other like previous generations of large language models, is that nobody spent like, I think it was like tens of thousands of hours or hundreds of thousands of hours into improving the model using like direct human interaction, because mostly it was like you take this data from the internet and you train this model kind of in a computational way and then you put it out there. Whereas they took you know months and months to actually hire these people and ask them to improve the model, like really on an interactional basis. And then for me, that was the big thing. And then no one really talked about it. They were like, oh, Chat GPT came out, it's the best language model ever, blah, blah, blah. So like so many things get swept under the rug.
SPEAKER_00Yeah, I feel like the first inklings I got of that were the time story. I think it was in January of 2023 that talked about the workers in Kenya. And then that like started to kind of get it into the conversation a bit more, but it was still like among people like us who are really paying attention to this rather than you know the much wider public and the broader discussion that's happening around these technologies.
SPEAKER_01Exactly. Like the level of misdirection that's happening is quite impressive. I mean, I guess it's kind of part of the narrative, but like I really feel like for each person who talks about like the actual cost, there's like 50 people who are like, this is gonna revolutionize whatever it is you do, it's gonna change it. You should be using these tools. Like, why aren't you doing it already?
SPEAKER_00Yeah, and as you were talking about that too, I was thinking, you know, have you mentioned that a lot of the discussion around this, that, you know, the company was really pushing out there because, you know, open AI, because I think it was beneficial to them, was this question of like, is generative AI gonna be this big threat to us because it might become intelligent and then, you know, this like threat to humanity if it wants to take us over and kill us? And it's like, what if the threat that we were talking about was like how much energy and compute that these things required and how much emissions that was going to generate as a result? But like, that's not the threat to humanity type of discussion that we want to have. It's more fun to have like this sci-fi scenario that also directs our attention somewhere away from the real kind of regulatory questions, potentially, that we should be looking at.
SPEAKER_01Well, actually, the like existential risk or the long-term risk link with climate change is really interesting because from that particular point of view, climate change is not an existential risk because it won't kill everybody, essentially. So it's like, I mean, quote unquote, it's okay because like rich global North people are probably gonna be able to like, you know, build barries or or like protect themselves or have bunkers or whatnot. My biggest frustration with this relationship between long-term risk or existential risk and climate change is that you know, you there it's not really considered an existential risk because like a lot of the global north and richer countries are technically gonna be fine. And so when you go on these websites, they say that you know the existential risk of climate change is close to none, close to like a non-existent compared to AI. So AI could technically like really literally wipe everyone out, but whereas climate change, like most of like the rich white people are gonna be fine. And that's just such a privileged take that just like really makes my skin crawl.
SPEAKER_00Oh, yeah, totally. I read William McCaskell's book and I was looking back over some of these things recently for something I was putting together. And there's even this thing that he has written about in the past. It was in a piece that Emile Torres wrote about, you know, all this stuff, where McCaskell was basically arguing that even 15 degrees of warming probably wouldn't eliminate all agriculture. So we as a species would be like, okay. And it's like, what are you talking about?
SPEAKER_01Exactly. Yeah. And so I think it really feeds into this whole discourse, right? We should only be focusing on AGI. We shouldn't be talking about climate change at all or AI's climate impacts. And every time I'm like, how is this even like logical, right? Like we're already going through like the hottest summer for so many people. Like we're already going through all of these things. How are we not focusing on the tangible like here and now? Why are we talking about this like years into the future potential global annihilation situation? I don't know, maybe I'm too like down to earth to really buy into this long-term risk stuff. But for me, it's like a no-brainer.
SPEAKER_00How dare you think about real life and not what's going to happen a million years from now?
SPEAKER_01But eventually gonna happen. And then you have people like Sam Altman who are like, oh, what we need to power AI is an energy revolution, and like I'm very conveniently investing in this nuclear energy company that's gonna come along and solve all these problems. So we shouldn't even be talking about the energy usage because this, like, this nuclear fusion or whatever startup is gonna just solve that anyway. So like there's so much like pushback to all this.
SPEAKER_00Yeah. And then if my nuclear energy company that I'm investing in can't come up with this energy revolution in time, then shrug, we'll just geoengineer the planet to make sure that we can row out all these AI tools, right?
SPEAKER_01Yeah, yeah, exactly. And there's always some like technological fix somewhere very soon, though, right? Like, because that's the thing, it's imminent. We're gonna fix this problem imminently, so let's just not even talk about it. And yeah, there's a lot of that in like especially like the tech industry circles or like the conferences and stuff that I go to. There's always like, why are we even talking about this? We should be talking about like streaming. How about like the environmental footprint of streaming or Bitcoin or whatever? Like, we should be talking about those, like AI is not an issue.
SPEAKER_00Yeah, and you know, we do talk about those, and those were bigger things when they were what was really driving it. But it's like generative AI is what's really driving it now, and that's why we're talking about it. And I feel like when you talk about those views that like you hear at conferences, it helps when they come from some of the most powerful and influential people in the tech industry, like Bill Gates, for example, who just a couple weeks ago was saying that governments should not be concerned with AI energy use, because when we develop good enough AI tools, they will just solve their own climate problems because they're gonna get like so good at this kind of stuff. And it's like, you know, again, like the techno fixes, but also like don't worry about what we're doing and specifically what Microsoft is doing, because Bill Gates is still very involved in Microsoft's AI strategy. Let us do whatever and the tech will solve itself, right?
SPEAKER_01Yeah, yeah, yeah. And I mean, it's true that there are some like real problems, like climate-related problems that AI can help with, but like the most successful use cases I've seen is really like when you take AI as part of a solution and a relatively smart, small part of the solution, and then the rest of it is like domain expertise, like people who actually know what they're doing here, and then they take AI and they kind of use it to improve their existing processes. But it's not like a, oh, we had no idea what we were doing before, and then AI came along and like solved the whole thing for us. Like what really bothers me is it's really this whole like perception of AI as like an actual tool, like as a standalone tool that will solve climate change, because that's just false. And you know, sometimes I uh when I talk about like environmental footprint, they're like, oh, well, AI is helping like improve climate modeling or whatever. And you're like, yeah, but that doesn't actually change the fact that like the energy demands are growing disproportionately to any kind of like efficiency gains you can get from AI in climate modeling, for example, right? It's just like it's two separate discussions, really.
SPEAKER_00There's nothing quite like the smell of fresh baked bread coming out of the oven. What if I told you that you could get the delicious experience of homemade bread with none of the time and work involved? Well, you can from Wild Grain. Wild Grain is the first ever Baked From Frozen subscription box for sourdough breads, fresh pastas, and artisanal pastries. Every item bakes from frozen in 25 minutes or less, no thawing required. I'm told on good authority that they're pretty delicious, and you can now fully customize your wild grain box, so you can choose any combination of breads, pastas, and pastries. You can even build a box of only breads, only pastas, or only pastries if you'd like. Plus, for a limited time, you can get $30 off the first box. Plus, free croissants in every box when you go to wildgrain.com slash save us to start your subscription. You heard me. Free croissants in every box and $30 off your first box when you go to wildgrain.com slash save us. That's wildgrain.com slash save us, or you can use promo code save us at checkout. I wanted to reposition our conversation to dig a bit more into this question of the types of models, right? Because I feel like this discussion that we're having is based on this general idea that we need these models that can try to do everything, that are, you know, are trained on like every possible bit of information that we can possibly access to create these kind of generalized models that, say, the OpenAIs and the Googles and Microsoft's and stuff are putting out into the world. But then there's also this other part of this where people are arguing that we don't need these general models. What we need is like smaller, tailored models to the specific use case that we're trying to do or to take care of. And those would actually work better while having less energy use or computational demand than these like general models that these major tech companies are pushing. Is there anything to that argument? What do you think of that?
SPEAKER_01So I think that the argument holds if you're like one of these companies like OpenAI or Google or Microsoft, eventually, like that you kind of have all these very generic tools like ChatGP, for example, it's like a tool that's technically like meant to do all these things. So maybe for a specific tool like that, if that's what you're trying to do, then having this general purpose model makes sense. But like the vast majority of applied AI use cases are very specific problems. Like I've worked in finance, I've worked in like automotive AI, like essentially it's like you need to do something, and then you're gonna train an AI model or use an existing model to do that particular thing. Like I've never heard of a like a client or whatever being like, we want to do everything. Like we need a model that does it all. And so I think that you know, if you're coming at it from an actual like user perspective, you need like very specific models, not only to like a task, but also often to like a specific domain. Like not only do you want to answer questions, you want to answer questions in finance. And it doesn't make sense to use this general purpose model because, for example, there are words that have specific meanings in fields like I don't know, whatever, bull and bear and whatnot. Like you want your model to be specifically catered towards a very specific use case. And so I think that this discourse of like, oh, you need these general purpose models is once again part of the marketing spiel around generative AI because at the end of the day, like these models have to be making money at some point. So you want to be telling people that they should be using them because they'll they'll pay for themselves, right? But if you really think about it, what you need in most cases are test specific models.
SPEAKER_00And how much is this push for these like general models that do everything? How much is that related to, on the one hand, the companies, you know, these major companies like the Googles and Microsoft's and whatnot, trying to create this, you know, situation or market dynamic where it's only major players like them who can really compete on the scale, versus how much is it kind of motivated by, as you were saying with the long-termist stuff, this desire to like try to create the machine that thinks like a human and is the artificial general intelligence and sort of, you know, those sorts of ideas. How much the both of those motivations kind of play into this?
SPEAKER_01That's a really great point because it's true that like in the 10 years that I've been working in AI, like the barrier to entry has shifted so significantly. Like at the beginning, like if you were a grad student, you could like train a model on your laptop and submit it to like one of the big conferences like NERIPS or or or one of the like, I don't know, like ICML or whatnot. And then, you know, if the math or like this computer science part was sound, you could get accepted nowadays. It's like not only do you have to compete with this like bigger is better paradigm in terms of like, oh, well, someone trained a 70 billion parameter model, so you should be training an 80 billion parameter one. There's also like this benchmarking aspect of like you have to run previous people's models in order to prove that yours is better. So the onus is on you to say, well, you know, I compared all of these, like you need the compute for that. Anyway, so like the like not only like now graduate students are having trouble, we're seeing more and more like academic institutions partnering up with big tech because they need those compute grants. They need that, like essentially those GPUs. And so it's like that particular dynamic has become kind of very skewed, and like a lot of researchers will have a double affiliation with academia and industry just to be able to compete because like essentially you need to publish papers, you need to write like and and then so it's yeah, so that part has become very weird. And also, as you said, like the companies that can afford to really train these models from scratch are such a select few that they're monopolizing the industry. And if it's like everyone wants to use an LLM and there's only a couple of organizations that train LLMs, then that's who you're gonna be going with. So I definitely think that there's like a concentration of power that's getting worse and worse, and the environmental impacts of that are part of that narrative, but not being transparent essentially.
SPEAKER_00Yeah, and you even see that with how these deals are playing out, right? Where a lot of open AI's deal with Microsoft is related to cloud credit so that it can use the Azure servers and stuff like that for its training and running its models and its services and all these sorts of things. And of course, any of the rising AI companies like Anthropic, for example, basically are required to get these massive investments from Google or Microsoft or Amazon or Meta to like compete in this space, right?
SPEAKER_01And then everybody is reliant upon NVIDIA because they have are the bottleneck of all of this. And if you even want to build your own, even assuming that you have the incredible amount of money needed in order to build your own cluster, you still need to get access to NVIDIA GPUs and there's a real bottleneck there and uh like a wait list and whatnot. And so also like that particular concentration of power, like there's literally one company that's making all the hardware for training AI models.
SPEAKER_00And and how does NVIDIA then play into this bigger picture, right? When we're talking about the climate costs and the environmental costs of making these major models, you need the graphics processing units from NVIDIA in order to do that. Or like, you know, it is kind of the major supplier and we've seen its share price like go crazy as a result of this whole moment. What's your thought on how NVIDIA, you know, kind of fits into this bigger picture?
SPEAKER_01I'm really frustrated because I've been talking to them for like years at this point and asking for more transparency because for like a lot of companies, even like Apple and Intel and whatnot, they create like life cycle assessments or carbon footprint assessments of their hardware. And it's, I can't say it's simple, but you know, it's straightforward. There are methodologies, like people do have been doing this for a while. And essentially it's like you take a look at your supply chain, you ask your suppliers to because the thing is, like, for example, Apple won't manufacture their hardware in-house as well. Neither does NVIDIA actually. They all always outsource this to their suppliers, right? And so essentially you ask your suppliers to quantify with like the lifecycle assessment methodology, like how much energy they're using, how much water, uh, like transportation, raw earth metals, blah, blah, blah. Like you have all of these, like, I mean, it's a well-defined methodology. And then once you have those numbers, you integrate that into your scope three emissions, that your supply chain emissions. And actually, like for Apple and Intel and stuff, for specific pieces of hardware, they will have a number in terms of like CO2 emissions. And you know, if you buy a, I don't know, whatever, a MacBook Pro, it's this much CO2. And so NVIDIA have been trying to get those numbers for them for for years. And I mean, they they say that the supply chain is complex, sure, as is the case for a lot of hardware. But the thing is that there's a single wire that makes all of all of NVIDIA GPUs. They're based in Taiwan. Um, and just like even if we don't have the raw numbers, uh, Taiwan's electricity is 100% coal-based, which is the most polluting like electricity type. So we know that that already is like the intense amount of energy needed to create this hardware is essentially a huge problem. Also, pure water, because when you're creating these like chips, you need to purify every tiny little layer of silicone, and now it's like nano nanometers of silicone essentially. And then so they're using crazy amounts of water. And to the extent uh there were some droughts in Taiwan in in recent years, and the government actually told farmers not to plant crops so that the water goes towards like the fabs, the fabrication of hardware, because it's become such a geopolitical issue. So there's energy, there's water, and then there's like the metals, like there's all sorts of rare earth metals that are mined in like terrible conditions, and also like the amount of earth that you have to move in order to get like one gram of uh a rare earth metal is like, you know, there's like it's like one ton for one gram, essentially, in some cases, and the labor conditions and the human rights conditions are terrible. So all of this, like we don't have any transparency on any of that, and people are buying like thousands of GPUs in order to participate in the AI rat race, and that's not getting accounted for anywhere.
SPEAKER_00That's so wild. I I honestly didn't even realize like the whole scale of that picture. And when you talk about, you know, Taiwan basically saying, like, okay, don't plant your crops because we need the water to go to like the chip manufacturers, it kind of doesn't surprise me when you think about how important it is for that, you know, small island country to like have this key industry and to make sure it's still working properly because it's really like, you know, it's part of its protection like from the West, right?
SPEAKER_01Exactly. And it's it's really a geopolitical issue, right? Like the US president signed an order like limiting how many, how many GPUs can be sold. I mean, there's all these like kind of geopolitical aspects, but also like nowadays, since like chips are such a core component, not only of AI, but of like cars and like TVs and everything, right? And so just like all of that is so complex. And the fact that like Taiwan is close to China and there's there's all this stuff to unpack there. But essentially for me, it's it's really the environmental side of things and the fact that you know there's no excuse of there's a such a complicated supply chain, it's one supplier. You can get the numbers from them, even if it can take some pushing, because maybe this is not something that they do inherently. But you know, if you're someone like if you're a company like NVIDIA, that essentially is the most expensive company in the world or whatever, as of recently. Um, so if you're that kind of company, you have enormous market pressure on whoever your suppliers are, and you can tell them, you know, give me the like life cycle assessment of the GPUs that you're making for me, or else, you know, whatever, and then they'll do it. Like, of course, it's not like there's some, there's there's a huge disbalance of power. So the fact that NVIDIA hasn't published a single carbon or life cycle assessment for me is a is really a glaring omission in the grand scheme of things when it comes to AI.
SPEAKER_00Yeah, and tells you they probably just don't want those numbers out there because they could get them if they wanted them.
SPEAKER_01Exactly.
SPEAKER_00And then, of course, when we have these discussions about the broader impacts and the climate impacts of generative AI, that's a piece of the picture that is not into this broader discussion that we're having or these figures that we're using because it's not transparent, it's difficult to get those numbers.
SPEAKER_01So a couple of years ago, I was working on the like the big science project, and that's actually how I joined my current employer, Hugging Face. So essentially it was a first time that a large language model was trained by uh like in a community way, and it brought together like a thousand researchers from around the world, and we got compute from like a public compute cluster in France. And then I was responsible for the carbon like accounting part of things. And then I started thinking, I was like, well, if you look at life cycle assessment, like if you buy a pair of jeans or a tote bag or anything, you can get the card, like the life cycle assessment that goes from you know the cotton that was used and how much water that was used and the transportation and like it will cover all of the steps of the life cycle, and that's how we typically think about products nowadays. It's cradle to the grave. And so for AI, that didn't really like no one really thought about it that way before because everyone was so focused on training. And so I we we did this life cycle assessment, and you know, we looked at not only like, for example, the GPU energy consumption, but also like the whole data center and the overhead and of all of that. And we found that if like just for energy, the GPU is only half of the overall energy used for training models. So any numbers we had until then, we could multiply by two, and those were the real numbers. And then when we started looking at the manufacturing process, like I that's when I reached out to NVIDIA, this was like three years ago. And then I realized that like they haven't published any numbers, and the only numbers we could get were from like the actual like data center like hardware, like the nuts and bolts of like network adapters and cables and all of that. Like you have that. It's it's like especially for a public compute cluster like the one we were using, they had those numbers because they're like kind of obliged to have them for transparency reasons. But for GPUs, there was not a single actual like validated number anywhere. And that really blew my mind.
SPEAKER_00That's so wild. I do want to go back to one other part of the question that I asked that we didn't get to, which is you know, we talked about the business angle of this and how much the companies want to make these large general models to try to reduce competition because they are the only ones who have the compute necessary to train them and to operate them and things like that. But then there's the other piece where a lot of these, you know, figures who are influential in this space talk a lot about artificial general intelligence and wanting to achieve, you know, the computer that thinks like a human, basically. Do you think that that is part of the motivation for pursuing these general models? Or does that just provide like an ideological or rhetorical justification for having done it?
SPEAKER_01I personally think that it's mostly providing like a distraction, also a way to avoid regulation or scrutiny, because it's like, well, if I'm single handedly building AGI, uh artificial general intelligence, then you know, like I'm well, like, first of all, it's like you can't really regulate me. I have all this power, like, and who cares how what the cost is if this is what we're achieving. And I think it's also a way of contributing towards like the oh, all these other other problems we'll solve. themselves, like the energy usage or the labor cost or whatever, because like once we have AGI, none of these things are going to be problems. And so I think it really is a like a contribution towards like shifting the focus. And also like if you really believe in this, like why would you want to stop someone from like solving all the world's problems, right? Like whatever it takes, they should have it. It's definitely really sad because like you hear a lot of that when you talk about those costs that essentially get swept under the rug, rug. When you when you bring them up, people are like, well, it doesn't like they kind of like dismiss it by saying, well, it doesn't matter because we're pursuing AGI. And now I feel that like before it was mostly kind of like more corporate marketing people, but now even researchers have this. Like when I before I was like ask people to be a bit more transparent when it comes to the compute cost of their papers and of like the stuff that they publish in the AI space. And now I ask them and they essentially have the same discourse saying, doesn't matter, we're working on AGI. And for me, as like a scientist, I find that really shocking because it's like, well, yeah, you have scientific responsibility. You should be transparent about the cost of your work. That's part of being a scientist, that's part of being a researcher. And the fact that it's percolated to that level, yeah, really worries me.
SPEAKER_00You wonder if like it's a really deeply held belief, or like you know it's the way to get funding is like to talk in this way.
SPEAKER_01Yeah, I guess it also like when you look at generally the state of transparency, like a couple of years ago, before ChatGPT, honestly, you'd have some form of like ballpark numbers. Like people would be like, oh, well, we trained on whatever, 2,000 hours on these kinds of GPUs or TPUs sometimes, like the Google tensor processing units, like they have their own hardware. Anyway, like they were relatively transparent. Like it wasn't necessarily like a secret. And nowadays, if you read these papers, especially like the more like industry lab ones, they have no information about how they train the model, where the data was coming from, how much compute they were used. All of this information just like doesn't make it out there anymore because of this like secrecy or rat race around Gen AI. And I feel that like from a fundamental perspective, that's problematic. Like that's not how you do science.
SPEAKER_00Yeah, exactly. It's all it's all trade secrets now, right? Nobody can know what we're doing because we also just don't want you to know.
SPEAKER_01Right, exactly. But like when you really like poke at it, like just by saying how much compute you used, it doesn't really give you any any insights about your model as such. Like that's what really bothers me. Like it's really kind of this, like I've worked in corporate settings. Yeah, you don't like there's this like keep them, feed them shit and keep them in the dark approach, but that's like a corporate approach. Whereas like, even if I gave you like a ballpark where my data was coming from and ballpark how big my model was and how much compute I used, you still could not be able to reverse engineer my large language model. That's impossible. You need the code, you need the actual data, like you need all these components. So the fact that like as a discipline we were going in that direction, like really shows that now the corporate narrative has fully taken over AI, at least in the large language model space, and that's like the direction we're heading.
SPEAKER_00Yeah, it it's very concerning, but but not wholly surprising. Like it makes me think of a few years ago when the dolls in Oregon was fighting Google to try to find out how much water its data centers they were using. And for a year it fought the city basically in court, the local publication that was trying to get these numbers from the city.
SPEAKER_01And they were fight funding the legal fees, yeah.
SPEAKER_00Exactly. Yeah, it was funding the legal fees of the city to not release the data, even though data centers nearby from Amazon and Apple were releasing those figures and we're like, we don't know what the problem is, but Google did not want to share it. And then after a year, when you know the kind of the public backlash and stuff got large enough, it finally said, okay, here's some numbers. But it was like, why don't you just tell us how much water you're using? Like it's not this big deal. It we're not gonna know this crazy thing about the data center and like your business practices because we know that. Like it should just be easily available information that the public has a right to know, not this like crazy thing that the companies are trying to hide just because they can.
SPEAKER_01Well, I think that also the problem is that like big tech has had a very relatively good climate reputation, an environmental reputation. Like, I mean, what's interesting is if you look at it from like a really like a global perspective, like Google and Microsoft are are some of the world's largest purchasers of renewable energy. I mean, like for years and years, like when you read the SG reports, I have a thing for reading the SG reports, and it's really like we're at the forefront, we're investing in this research. They have tons of people working on like climate modeling and things like that. So I mean, like there has been this emphasis on the positive side of things. And so I think that when it comes to actually opening up their books and showing us the cost of all that, that kind of potentially questions that narrative of, oh, we're like doing all this good work to stop climate change or to protect our planet. And I think that's where uh things become complicated. It's like when you actually ask the cost of all that. And and like in general, I feel it nowadays it's like from different angles, whereas it environment or labor, whatever, like the cost of AI to a society is really not part of the discourse. It's really like that the emphasis is on the benefits and not the costs. And that's not a good way to think about innovation, right?
SPEAKER_00Yeah, it's interesting, right? Because I feel like the places where that starts to seep into the conversation is when you have these moments where these communities start pushing back against these data centers, especially in this moment where the major companies are, you know, continuing to and accelerating this build-out of data centers around the world. I was really struck by a figure that Microsoft gave earlier this year when they released in a report that they had five gigawatts of installed server capacity at the beginning of this year. But in the first half of the year, they wanted to add an additional gigawatt on top of that. And in the first half of 2025, they wanted to add another 1.5 gigawatts on top of that. So this like acceleration of what they have versus what they're adding in each new, say, six-month period or whatever is just like continuing to scale up. And you have to wonder on the one hand, do they really need that much server capacity? But then on the other hand, like, is there going to be this much demand once the hype around generative AI kind of starts to fall again? And what happens there? I feel like it's hard to understand. But the one thing that's very clear is that their energy demands, their computational demands, their water demands are not going anywhere and are only going to continue rising in the years to come.
SPEAKER_01Exactly. And also as a provider of services, like they are also essentially the way I've been I've been trying to like present this to people is that instead of seeing AI as like a vertical, so if you talk to like climate folks, like they tend to think in terms of verticals, like the IPCC will talk about agriculture and transportation. And like you have these verticals that have been kind of part of the discourse for for decades now. And AI is typically put in the ICT vertical, like the information and communication technology vertical with phones and internet and stuff like that. But I think that it's more of a horizontal and it can impact agriculture, it can impact transportation, like that's the thing. And then we don't really know what the impact will be. Like maybe, like, maybe it will make some things more efficient and make the emissions of the whole sector go down as a result of that. But I think what we're seeing so far is that it's actually like contributing towards amplifying whatever the existing emissions of that sector are in many cases. So you're making something more efficient, but people are using more of it just because now generative AI is in all the tools. Like they were going to integrate generative AI into Google Maps. And like how many people use Google Maps, right? And so it's like it will make the overall emissions of that sector, which includes like the navigation part of it, bigger because now we have Gen AI.
SPEAKER_00I have a couple final questions to close this off. I wonder on the personal side of things, do you think that there's anything people should be kind of keeping in mind when they think about generative AI and interacting with these sort of tools? Or do you think it's a bit more of like a question of what these major corporations are doing versus what the individual is is doing and you know, interacting with these things?
SPEAKER_01I generally feel that in the context of climate change, we've put a lot of pressure and emphasis on individual action, which is definitely part of it. But you know, like the concept of carbon footprints was invented by British Petroleum, like it was a way to kind of shift the responsibility on individuals and not on like oil and gas companies, for example. And I think that with AI, it's a little bit similar. Like we, of course, can have some level of agency and we don't need to use Chat GPT as a calculator, for example, but it's unfair to tell people to, you know, stop using AI because of its climate impacts. I think it's really like the pressure or the responsibility should be put more on the providers of these tools, for example, in order to provide the energy or the carbon that's linked to a like, for example, a Google search or an interaction with Chat GPT, right? Like, and in order for people to be able to make their decisions based on that, but I think like the first point of action should really be on providers and on the people like making these technologies, because it's like if you have existing structures where, for example, if you're using living in a rural place and you don't have public transportation, you really have little choice than to buy a car. It's kind of like similar with AI, I feel. It's like if you're already in these structures and every time you do a Google search, it uses generative AI, like you're kind of like boxed into this already. I mean, of course, you can use Ecosia, for example, which is a search engine that is actually carbon neutral. But other than that, like you're like, you know, you're using Gmail as part of your job and you're using whatever as part of your personal life. And so you're already kind of like stuck in these systems. And so I would put the pressure more on the on the providers of the systems than the individuals themselves.
SPEAKER_00Yeah, I definitely agree with you. And you know, I think listeners of the show will know that there are certain like tech products that I don't use. Like I, you know, I don't use Uber and I don't use the food delivery apps, and I've actually never used Chat GPT like to go onto the website and use it. But I definitely don't think I'm like saving the world by doing that or anything, you know, um, because ultimately the question is what these companies are doing and how they're being forced to change on a broader like structural level versus what us as individuals are are doing, right? And so I guess the flip side of that then is, you know, we talk about individuals and that this isn't inherently an individual issue. So then when you're looking at what these companies are doing, as you were saying for a long time, I think it was fair to say that these tech companies did develop a reputation for being more environmentally conscious or climate conscious than some of the other major companies out there. And I think that they put a lot of emphasis on trying to make sure that they were seen that way, you know, by say making pledges to be net zero or carbon neutral, by buying a lot of renewable energy, by making sure they were buying carbon credits to say that, you know, they were kind of replacing their emissions or whatever. And obviously there are big questions about carbon credit and offset schemes and things like that. But what is seeing their response to this AI moment and how quickly those things have been kind of pushed to the side and how eager they have been to kind of chase after this regardless of the impacts? What has that told you or what have you been thinking about their commitment to climate action as a result of what has been happening over the past year or a couple of years?
SPEAKER_01I mean, I think that capitalism in general is at odds with sustainability. And it's just like, I mean, if you think about it, at the end of the day, the point of a company is to make profit for its shareholders or whatnot. And so when it comes to trade-off between sustainability and profit, profit, it's like why companies exist and like CEOs have a responsibility towards the board or whatnot. Like, I mean, it's it's kind of that's the mechanism, and so you can't really expect them to choose sustainability. But what I do think is that, I mean, maybe I'm a little bit naive, but I do think that like this and the energy and the climate impacts of generative AI actually did spiral in a way that even the tech companies didn't expect. And now the question is like, what now? And and I think there's gonna be increased tension between profit and sustainability around AI. But what's interesting is that like currently I don't see the massive improvement that Gen AI brings. Like, I mean, I see it being sold as such, but I don't really like tangibly see the big deal, quote unquote. And so I think that maybe as you say, like the AI summer will will die down a bit when we realize that okay, maybe energy is gonna be part of the conversation, but like that it's not living up to the expectations that, yeah, Chat GPT like answers questions, but it also hallucinates, and you know, and you know, there's there's essentially so few use cases where it's actually, I guess, like irreplaceable. Then I think that that's when kind of like it's gonna be like too expensive to maintain it than to keep using it. And so maybe this is kind of like a more of like a capital versus environment struggle, and it's nothing new, honestly. It's like it's part of the last couple of decades, right? Of lack of progress around climate change because money.
SPEAKER_00I think that makes a lot of sense though, right? Because I think that there's often this idea that generative AI is inevitable, right? That it has to be rolled out everywhere, that we have to use it. And I tend to think of like, you know, going back to the metaverse moment, like there was this push from Meta and from some other companies to do this kind of project that was going to be very computationally intensive by having us all spend a lot more time in these like 3D environments. And I think we headed off a real climate disaster with like defeating that admittedly very bad idea, which was probably not going to go anywhere anyway. But I think that based on what you described there, what I hear from it is that generative AI is not inevitable. There might be some use cases where it makes sense to be using tools like this, but the idea that it's going to be rolled out everywhere and that we just need to get used to this is very much something that is not baked in right now. And there's still the opportunity to change course and say, listen, it's okay to use these things in certain instances, but it makes no sense to build them into Google search and to build them into these like infrastructures of the web that we're used to because they don't provide the benefits.
SPEAKER_01Yeah. And for example, in France, I keep hearing this term of digital sobriety, which I find so refreshing. Like I've been participating in a couple of discussions or like events where people are literally like, do you really need a new cell phone? Do you really need to be using ChatGPT? And I feel that it's questions that we don't really ask ourselves that much, maybe because like there's more pressure, there's more marketing or whatnot. But it's already started in Europe to some extent, this pushback against generative AI or AI in general as the solution to every problem. And I find that so refreshing because I'm not hearing a lot of that in North America.
SPEAKER_00I love that. Let's all commit to digital sobriety. And, you know, whether you have sobriety uh, you know, in in your real life, what you're drinking, that's totally up to you. But let's uh let's at least do our digital part. Um, Sasha, it's been wonderful to talk to you. It's, you know, been so enlightening to learn more about this whole space and what's going on with it. Thanks so much for taking the time.
SPEAKER_01So great to talk to you too.
SPEAKER_00Sasha Lucioni is an artificial intelligence researcher and climate lead at Hugging Face. Tech Won's Save Us is made in partnership with The Nation magazine and is hosted by me, Paris Marks. Production is by Eric Wickham, and transcripts are by Bridget Pelou Fry. Tech Won't Save Us relies on the support of listeners like you to keep providing critical perspectives on the tech industry. You can join hundreds of other supporters by going to patreon.com slash tech won't save us and making a pledge of your own. Thanks for listening and make sure to come back next week.