Sustainable AI with Kacy Clarke

Welcome to the Artificial Intelligence Podcast with your host, Jonathan Green! In this riveting episode, we dive into the sustainability of AI with our special guest, Kacy Clarke, a seasoned IT strategist and enterprise architect.

Kacy provides an insightful perspective on the burgeoning energy demands of AI, highlighting the need for a sustainable approach to balance the power and cooling requirements of data centers. She discusses the critical intersection between AI's potential and the real-world constraints of energy resources, emphasizing innovative solutions in nuclear energy and the development of small modular reactors.

Notable Quotes:

• "It's a huge crisis... we're already slowing down data center builds because of the power consumption." - [Kacy Clarke] 
• "Even though it's inside of a computer, it still affects the world." - [Jonathan Green] "I don't need a large language model that has understood all of human writing... let me talk to the expert." - [Kacy Clarke] 
• "The biggest nuclear power plant in the US generates about 4.6 gigawatts, and we just don't have enough energy." - [Kacy Clarke] 

Connect with Kacy Clarke:

LinkedIn: https://www.linkedin.com/in/kacyclarke/
Website: https://sustainablearchitectures.org
Chiefly & Co: https://www.chiefly-co.com/

Connect with Jonathan Green

• The Bestseller: ChatGPT Profits
• Free Gift: The Master Prompt for ChatGPT
• Free Book on Amazon: Fire Your Boss
• Podcast Website: https://artificialintelligencepod.com/
• Subscribe, Rate, and Review: https://artificialintelligencepod.com/itunes
• Video Episodes: https://www.youtube.com/@ArtificialIntelligencePodcast

Transcript

0:00

Is AI sustainable with today's special guest, Casey Clark Today's episode is brought to you by the bestseller Chat, GPT Profits. This book is the Missing Instruction Manual to get you up and running with chat g bt in a matter of minutes as a special gift. You can get it absolutely free@artificialintelligencepod.com slash gift, or at the link right below this episode. Make sure to grab your copy before it goes back up to full price. Are you tired of dealing with your boss? Do you feel underpaid and underappreciated? If you wanna make it online, fire your boss and start living your retirement dreams now. Then you can come to the right place. Welcome to the Artificial Intelligence Podcast. You will learn how to use artificial intelligence to open new revenue streams and make money while you sleep. Presented live from a tropical island in the South Pacific by bestselling author Jonathan Green. Now here's your host. Now. I'm so excited to have you here today, Casey, because for me, this is a really interesting issue. I actually just saw on the news today that they think it's gonna cost $1 trillion in electricity data centers and communications just to meet the promises that AI has already made. And if there's one thing I know, it's that next month AI's gonna make bigger promises than it did this month. So we're hitting this. Wall between promises and what's possible because sometimes we wanna go to the moon a million times, but then it's how much fuel is there? So there's these other structural things that we skip over that are critical to that success, and you can't have artificial intelligence, electricity, and you can't have electricity without something to make the electricity. So I'd love to start from there. What you think is the right direction, where we're shifting wrong already, like right out the gate. So I think right outta the gate, the $1 trillion actually came from Sam Melton the CEO of OpenAI. And he, along with a number of other leaders, went into the White House in September. And OpenAI is proposing that they need in order to get to the, $10 trillion parameters are currently over a trillion, which at GBT four, you know, or a hundred trillion, or however many it is that they're going to need for their trajectory in the next few years to build five to seven five gigawatt data centers. So five gigawatts is enough to run Miami. And at 30 million square feet, and most of the new GPUs require water cooling. So it's, not only the energy, but the water's going to take, because air cooling is no longer sufficient with how much heat all of the GPUs and TPUs are taking. Five gigawatts. Currently the biggest nuclear power plant in the US is right outside. In Atlanta, it's in Georgia. It has four reactors. It generates about 4.6 gigawatts, and we just don't have enough energy. They're keeping coal fire plants coalfire power plants online in Northern Virginia right now because they can't take 'em off. And feed, the hundreds of data centers that are in Northern Virginia. So this is a huge crisis. They're already slowing down. Data center builds in places like the Netherlands and Ireland because of the power consumption that's coming in as well as the land that's being used. And we're gonna have to find a balance between all the good. And it does do a ton of good that AI provides us and real sustainability goals that, the cloud providers have pitched in and we've committed to in the Paris Court and other types of agreements. So one of the things I find really interesting is that like I'm someone who grew up on the Terminator movies. We think about the robots killing off everyone because they think we're inefficient, but maybe they'll just take all the water to, because they need it to cool down. And we already have a lot of places where there are, they're limiting water to the farmers because of large corporations or these places. And we are starting to. Like food prices are going up and maybe this is why, maybe we need to be a little more cautious and glow, grow a little slower so that we don't accidentally break down our food supply chain. Because a cascade there could be devastating and it's really amazing how, yes, AI is exciting and I love ai, use ai, but I like eating more, I'll be honest with you. If those are my choices, eat today or play with chat tea, I'm gonna eat. But we've created this. Really interesting situation because we've been heavy on conservatism so for so long, we haven't built a nuclear power plant in America in I think since the early eighties. I was looking it up since shortly after the three mile aisle incident, which most people in most people's minds, three mile island is of the same significance of Chernobyl, except for what's really interesting is nobody got even got sick or injured from three mile Island. It was just something almost happened and we right. It's really a, it's actually a testament to how good our systems were that like the backups went in place and nobody got hurt versus what happened in Ukraine with Chernobyl. But we remember them as like equivalent events, even though one's called an incident.'cause I guess that's the worst you can call it. When nobody gets hurt, you can't go beyond incident. And the other one's a definitely a disaster because millions of people got hurt. Yeah. Became very trepidatious or very anti-nuclear power for a long time. I think it's in western Germany, there's a nuclear power plant, or maybe it's in Austria. They built this massive nuclear power plant, and then the people voted against nuclear power, so they never turned it on after spending billions of dollars building it. And they shoot a lot of movies there, which is how I saw it. So now we're suddenly I don't want nuclear power unless it's for ai, and then I'm okay with it. So we're seeing a lot of shifts in how people are okay with something we haven't been okay with for a long time. Like nuclear power has been. Pretty negatively perceived for a long time in America, and now suddenly because of ai, it's like back in vogue it seems like. We haven't built any new plants, at least like we stopped building major power plants. That's not actually true. We are building small modular reactors right now. And in fact, three Mile Island not the reactor that melted down, but the other actor was online until five years ago, and Microsoft just cut a deal with Constellation Energy to turn it back on with Microsoft as its only con customer. And this is true, there's a bunch of different nuclear power plants that are being negotiated to turn back on that were mothball before because people were afraid of it. And we also haven't figured out how to deal with the spent fuel rods. From nuclear power plants, but small modular reactors are actually on the upswing right now. They can go up. New scale is the only one that has a, an approved reactor from the NRC, the nuclear Regulatory Commission. And there's a bunch of projects going on in West Texas right now. With these SMRs there's other experience experiments going on. France has been running like 80% of their power off on nuclear reactors for decades. So they have much more modern technology and they're continuing to build them out there. On the other hand, Japan, after their accident, it's trying to get rid of all of them, but that means they're going back to fossil fuels. Both oil and natural gas and, it really is a trade off. Do we want all the power of ai, which is going to come with all the power and the cooling, or, do we need to make some better decisions around how we do ai? With things like, small language models, ones that are agentic ai, things that are much more efficient and there's great work going on in a number of different areas on s SLMs. The Green Software Foundation has got some great practices out there for how you run AI jobs and how you make choices and thanks, for example, which region. You're going to be running it in the cloud providers, 'cause there's a real big difference between trying to run it in Northern Virginia or trying to run it in Eastern Canada, where it's pretty much all hydropower. So one of the other things that to keep in mind is how's this gonna affect what electricity costs for everyone else?'cause there's one direction where they go, you know what? We're using so much for ai, we'll just make electricity free for everyone. Which I think would be a pretty smart move, right? Give 'em a lot of good positive pr. Or there's the other version where they're the price. They know that these AI companies who pay so much for electricity that now everyone's bill goes up. So 'cause in which case, that will be devastating for most Americans. So for most people around the world is that if the electricity prices are controlled by these large corporations, it could be a really bad. Result for people. And I think these are a lot of things no one's thinking about, which is that it's very exciting for use of AI and we don't think about electronic conservatism very much. I'll give you an example. I think about it all the time because I just got the power bill two days ago or I got the water bill and I was like, what have you guys been drinking? So we don't, when you pay the bill, that's when you think about it, but it's often, there's all these little things that add up that we don't think about until we start. And when you're a kid, like my kids leave the fri, we will stand there in front the refrigerator and just look inside, even though there's a smaller window they can look in. Now my refrigerator came with one of these new features where you can look in a little window to see what's in there. They still open the thing and just stand there for 15 minutes. And now I've turned into my parents where I'm like, shut the door. You're just melting money. Like we always turn into our parents and we, really don't think about the consequence of our behavior with ai. So we're so excited. For example, we don't think, is there a difference between using chat GB two, 3.5 and four and V zero one or stra any of the new strawberry variants that they're releasing to some people, but they cost significantly different amounts of electricity. So the biggest complaint people have. Lot of this is my fault because I wrote the paper on this. So there's a theory out there, which is called chain of Thought, which is basically chain of thought is an answer to ChatGPT trying to save electricity. So the way ChatGPT was designed is to give you the fastest answer possible because that uses the least amount of electricity every second. It's thinking. Costs a certain amount of unit of electrics, let's call it a unit. So if it has to give you, if it takes 10 thoughts, 10 units to get a better answer, they would rather get it in three. If they can't, a pretty good answer in three is way better than a perfect answer in 10. So chain of thought is like now, why? Now? When it gives you an answer, it tells you how many seconds it spent. Thinking I was thinking for 16 seconds I was thanking for nine seconds. And those are really. Big costs that are, because of the number of uses, the number of questions per day actually are exponential. So if people just switch to a model that uses twice as much electricity per day, it doesn't feel like you're doing anything different, but you're actually making a massive worldwide shift in the electricity usage. And I really think that it's, there's two big guilty parties here. I think the first one is the AI companies, because they're. What they should do is run every question through a front filter, which decides which model it goes to. Because if I say to you, guess which model can is the right one for your question, the odds of you getting it right are not very high. Maybe I couldn't tell you as a heavy user which questions I should send a 3.5 in which I should send a four in which I should send a one. I just couldn't tell. I can't guess. So what do I then do? Try it with both. So now it's even worse. So it's, that's something that I think would be an easy solve, right? To just have a filter. It's a switchboard. When you think about phone calls, I guess a hundred years ago when there'd be someone connecting the two wires, if they would just put that in front of the ai, that would do a massive decrease. The other thing is that we. We've become in a patient society. We now watch shorter and shorter videos. Nobody is gonna wait seven seconds, right? Which is the length of a TikTok video. They want the video to be over in seconds, seven seconds. They're not gonna wait seven seconds for a response. And because of that, the machines have to be always on. So we've created something where there's this always on necessity, and I think about, and maybe this isn't exactly right, but difference between ram and hard drive, which is like frequently asked questions should just be in the front of the AI's memory. Who goes, oh, someone's already asked this. Here's the answer. Because it could do a little short cutting there if it's'cause people ask the same. Question all the time. I don't wanna say dumb question, but people ask a lot of the same question like, who won the football game last year? You don't have to re go through this massive thing if it's just pulling up a fact. So that's where I think they need to separate these i, these different parts where it's just give me an answer versus do active research versus do active thinking. So those are two areas where I think they're so busy trying to make an AI that's sentient. Why not just take the one we have now and make it efficient? I completely agree with you and choosing whether you're going to use 3.5, which has about 175 billion parameters in it, versus four oh, which has around 1.7 trillion parameters is a massively. Different amount of energy. For those, two things, including, once you start including things like, all the training, the pre-training, the self supervised training, the re reinforcement learning with human feedback, all the different layers to do this. Four O is 10 x more expensive. Then running 3.5. So having those smaller, I consider it calling the experts. I don't need a large language model that has understood, all of human writing since the birth of time, which I know Sam Alman wants to do. I instead, let me talk to the expert who actually understands how to do this chip design, in, in CAD and how, or is this the right answer for, maybe I'm doing a coding problem. There's some of the things that, claude does much better in terms of coding and some things that chat GBT does better. How do you combine these together to be more effective? So small language models, agentic ai, modular ones there are many things we can do to make AI much greener and actually get more effective answers. Now you were talking about this new chain of thought reasoning, which believe it or not, my master's degree is in AI from decades ago. And it's really evolved a ton since then. But the chain of thought, which is chat GBT oh one, it's currently oh one preview is the one that's out there as being experiment with Silicon Valley. What I like about that is, just like you said, you don't have to do all the prompts and tell it the chain of thought. It's now being, for things like math problems and reasoning problems, they actually fed it much better data. Things like math and science, white papers and the quality of data in order to train. Oh and oh one preview is so much better that the answers coming out of it are actually gonna be much better as well. And that reasoning aspect of how do you break a problem down into pieces and follow it so that it's much more efficient. I think it's an exciting time, but again, it's taken a ton of energy in order to train that. What's the right answer here? So one of the issues is how chat GBT kind of approaches, they call it mixture of agents or like mixture of experts. So there's multiple experts in there. But rather than being firewall, it's like someone with split personalities. So even though you're talking to the physics expert, you're still talking to the whole person. Even when I use A GPT to narrow down its focus and say, just be an expert in this. It's still accessing the whole thing, even though I'm trying to limit it. And it's wasting huge amounts of data. And I see this on large and small scale. I see this on people who build local AI models for small projects. They still use way more than they need. And it's very common 'cause this is an area where there's a lot of confusion or maybe there's just not a lot of like good training, but it's, I probably, even though I'm a power user, I'm one of the people who uses and test things all the time. I've probably used. 5% at most of the different areas, right? I probably, I've never done anything in music. I've never done anything in these, any of the languages. There's so many parts of chat GBT that they're there that I don't need access to. And if GPTs actually worked in the sense that they actually did firewall, and even if it was a tier two, it goes, it will just try to use a smaller database for question if it has to, then it goes to the bigger one. Even then it would save. Like a 90% drop probably in electricity costs. So it's really, I don't know why I shouldn't be the one thinking of these ideas. Like it should be just doing it so simple since they're the one paying these massive bills. Maybe that trillion could be half a trillion or just a hundred billion in electricity. But it comes down to we're so busy chasing excitement. We're not paying attention to what's happening along the way. I see this a lot with, I can only imagine what the video generation is costing electricity, and every single AI video that I've ever seen has been terrible. Some of them have just been trash, but most have been terrible. I've never seen one. I would've watched if we didn't first say, this is an AI video, and people still talk about soa, which is the OpenAI video that they announced like a year ago. And why did they announce it? Because they were trying to take control of a news cycle.'cause there's a negative news story the day before. So they dropped the Sora release but no one's seen hide nor ha in a year from it. And we see everyone else posting Sora, killer Sora Killers. Everyone loves to talk. I'm like, it's even out. What are you talking about? Stop wasting your time. AI video is so much more electricity intensive in ai, music generation. These things that pull in a lot of other. Areas, but they're less useful. So they're more expensive and less useful. So I think that we're not, we're so busy talking about the panacea, that we're not talking about the process of getting there. Like it's really exciting to talk about going to the moon, but then it's how do you make a rocket? What type of fuel to use? Where do you get the fuel from? Talking about that's not exciting. Nobody wants to talk about where rocket fuel comes from or how it's made.'cause that's not exciting, but. It's a really important part of the process, and I think that's what's happening here. And what I, the thing I see that's really interesting, and I think this is a really big problem, is that the definition of AI has become almost meaningless because when I was younger, AI meant sentient robot. Really sim, right? Or it meant self-aware, sentient robot E. And it was either an a good or bad overlord, right? But that was really, it was like a very clear definition and then it shifted so much. Now when I talk to people, AI means anything you can do inside of a computer, it's pretty much considered to be ai. So a lot of people are using chat, GBT, they'll use chat, GBT, the latest model to do. Math, like simple calculator problems because it comes your default. Go to, I've seen this happen, or use it for really simple translations. And we are because once it becomes your habit, it becomes your habit. And we are not thinking about, we're thinking about, oh, this is the tool to replace everything. So a pocket knife is really useful, but you can get a pocket knife that has a hammer in it. It's not gonna be as good as a regular just hammer, hammer. It will net because the price of having multiple things in there is that it's okay at everything, but not great at any one thing. So we're starting to think of it as like a magic, pocket knife that can do everything, but it really isn't. So it's the other tools we still have in our toolkits are still useful, but I think we're seeing this mindset shift, which is really. Conf. A lot of people are confused and it's because of AI companies, press releases are very confusing. They're saying, oh, don't use a calculator ever again. What are you a bad parent? Your kids should be doing all their homework with chat GBT, and it increases your user count, which is what they want. But it also creates a lot of bad habits and we go through these cycles of bad habits where we. Like when I was a kid and you were on the freeway after you ate your fast food, you threw it out the window and you just, as a kid, I was like, wow, that's magic. It disappears as an adult now and a lot of other countries are still in that phase. You go through that phase before you go through a phase of you dunno what the right word is of littering before you enter a phase of conservative where you realize, oh my gosh. It actually, I throw something out the window and I come back a week later. It's still there. Like we had to go through that phase in America and then we, passed a lot of like laws about throwing stuff out the window on the highway, and it's really a pretty big fine now. And that's how you stop it. And now very few people would even consider doing it. Like even describing it like you're look on your face, what are you a monster? It's exactly like when I was a kid, it wasn't, I think the same thing, but when I was a kid many years ago, we didn't think that way. So I wonder how long it will take us to shift to this mindset of oh. Even though it's inside of a computer, it still affects the world. In fact, I was. In thinking about this problem, digital litter is a real thing. You know that this seductive lure of abundance out there, and I've been working in cloud computing for 14 years. I'm a IT strategist and an enterprise architect, and I've helped a ton of different companies move out of their data centers and move those workloads into the cloud. And now we have a problem where it's just growing bigger and bigger. On average, most major enterprises spend two and a half times what they were expecting in the cloud than. You know what they're actually doing. And it's because of this problem of digital litter is nobody ever gets rid of anything. There's this feeling that the cloud is infinite and we can just keep adding all of our own videos. The comparison was, I have a security around my house in Northern New Hampshire and I have videos of deer and skunks and, various other wildlife, the bears, and not deleting it. It's just out there. Do I actually need that? No, it's digital litter. And on the enterprise scale, it's much, much bigger. And a lot of it is sitting on solid state storage. Not even, diss, terribly expensive. Use a tons of ton of power. You've got S3 storage out there, there's five copies of it, you might be replicating it, certainly you're backing it up, to another region. It's digital litter and it's just growing massively. So it isn't AI as the only one. It's this belief that electricity and computing is infinitely abundant and not really looking. The fact that everything has to run on hardware, it has to be cool. That has to have, use energy. Oh, do you think the solution is. Like education based or software based. So there was a cloud company, maybe this is 10 or 20 years ago now, that was like when you upload a file, we'll look to see if someone else has uploaded the same file. Yeah. So that we don't have two copies of it. And I was like, that's very clever if you can make it work. I don't think they were able to do it because there is also, we have to anonymize it and do that part. They were trying to do two parts. So that's one thing where you control duplicates or. I went through this with my own company. I was going through my, one of my cloud servers, and I was like, why is my bill so high? And it was whenever we, I would do a YouTube video, so I record the raw videos and upload those. Then the next person does the first round of edits, the second round of edits, and they all end up in different folders. And this particular cloud provider wouldn't give them the ability to delete. My raw files, once the final version done, once I've published the video to YouTube, I don't need the raws anymore. I'm not gonna reuse the raw footage from two years ago. I just am not with my industry, but if you don't manually do it, nothing happens. And they don't really have, they don't 'cause data centers, like they don't care as long as they're getting paid. They don't really have a policy for oh, you haven't looked at this file in so long. We pushed it to cold storage or whatever. So that's one way to solve it. Another way is like. Looking for duplicates or looking replica or creating some type of system, and then just training your employees to treat it a little more seriously. So when I worked in. It Way back in 1999, back in the previous millennium, this was when the entire company I worked for would at the end of every day do a tape backup, and I'm thinking it was 30 gigabytes, I think it was that it was like a, what we consider now like a crazy for the entire building. So everyone's hard drive would all get scanned every night, put on a tape drive, then mailed off to Iron Mountain. We call like iron. Yep. There. It's Iron Mountain. Iron Mountain. So he said it every day and then we had a problem. I was the bottom guy at the IT to pole and the head of it was like, Hey, we're going over the limit on our tapes. You need to go through everyone's computer, figure out who's doing it. So we actually had to manually go computer by computer. We couldn't even do it from the central. I'd have to look on everyone's computer and see what programs they have and look at the file sizes to figure out who was being naughty and who was being nice. And there were just a few people being enough naughty that it was with games going over the limit or video files. And it seems like we haven't improved the way of dealing with these, this, these files or file organization. And even now on your own computer, like finding files, duplicates of files, file, organization. Like the file organization side of computers is so bad. It feels like it's just this leftover technology from a hundred years ago that no one's figured out a better way to do. To sort our files better. We can find stuff and organize them and keep from having duplicates or be able to find things. And I probably have two of every single file on my computer, so I'm not innocent here. I'm absolutely guilty. Oh, so maybe Totally. It's all connected because we're seeing like hard drive sizes constantly increase. So in when I was, when I, my first computer had no hard drive. The thought of a hard drive, like I couldn't even imagine that now we have hard drives that are bigger than all of the like, hard drives that existed in the world 20 years ago. Like we're growing hard drive sizes so fast, and then now it's I have an external hard drive that when I move data to it, it's so fast. I don't believe it. I think I didn't do it because I'm used to watching the meter and taking four hours or three hours, and I finally built one of my own, like 40 gigabytes per second one. I was like, that can't be real. That number's, it doesn't make sense. It's impossible, and I'll just watch a file and it'll happen so fast. I can't see it. So we, because hard drives went fast and because clouds just seem magical, we do just put so much stuff up there and we, I have files on my Dropbox folder that I haven't looked at in five years or seven years. And what we've all become is digital hoarders, is that we just keep, you know what, hoarders always pollution out there that nobody's gonna look at, nobody wants to see ever again. And if you and I are doing that as individuals. Imagine doing it. If you're some of the largest banks in the world or insurance companies with claims that go back forever it's just a massive amount and they keep so many different copies around it, and it's, what do you need, a backup of a system from two years ago, nobody's gonna go back that far. But yet, they think because of regulations, they have to keep all of this there and they have a, certain industries, it's seven years of records. Other industries, it's 10 years of records, which is why Iron Mountain, by the way, is actually doing a ton of stuff on sustainability. Kudos to those guys over there, but it, why do we need 10 years? Now I, yeah, I want 10 years of my medical records, but this is like so much companies end up taking, keeping 10 years or seven years of everything, all these log backups and stuff, and they don't have even have the systems to restore it. It's like, why? How can we get more narrow on our regulation? So it's actually only keeping the stuff that we need to go back and look at again. The rest of it is just wasting computer power. I think it's just in case. So whenever I actually, I used to work on, when I was in college as a volunteer in EMT, we went into a hoarding house one time and it is, whoa, like I didn't. This is before that TV shows or anything. And like you had to walk through these tiny aisles and it was just stacks of stuff. And in the middle of a stack would be an animal cage with an animal inside. So it'd be like magazine. Skunk Magazine. Like the skunks not getting outta that cage'cause it's just buried. And then it would be like at different heights, there'd be different animals mixed in with other just random stuff. And you'd say, what is all this stuff for? It's just in case. And when have you ever had an emergency where you've needed a broken hanger? It's the same thing and. Here's what I think about is that there's this movie from, I think it's 1928, called Metropolis, where we lost the original files. They have part of it, so the version you can watch now isn't the actual original movie 'cause they don't have all of the reels. And they did some things with special effects. I watched a special effects thing about it that like are amazing that you can't, people now don't know how to do it. Like special effects 1928.'cause they had to shoot it all real. So they would use like mirrors and six different sets at the same time to create multi-part scenes. Very interesting. And everyone thinks. Their family vacation photos from 24 years ago are metropolis. I think that's what's happening is that we think everything we have is for posterity like I am. The number of people who think about what if aliens find my DNA in a thousand years? Have I left enough information that they could recreate? Let's not worry about that. But a lot of people I think are thinking like that. It's that idea of I'm so important that in a thousand years they're gonna wanna find a piece of my DNA or sum me on the, I need to even love data in different places that they can recreate my memories and. Maybe that's the wrong mindset, but I think it all, I think as we talk about this, I think it all does come down to our approach to information, our approach to computers, our approach to, we don't really think about the value of data. I don't I. Sell my old computers. Now I, every once in a while do have to go find an old file or you have to find a receipt.'cause some companies like you have to find your receipt email from nine years ago. So sometimes, every once in a while one of those things does come up. So it's not, never, unfortunately we have to try and turn on the old computer, but my wife is let's sell your laptop from 12 years ago. I'm like, are you crazy? What if there's something in there that we need and. It's happened. The problem is it's happened two or three times in the last 12 years. Where we actually have needed to go into the old laptop, find a file here or there, or you have to find a registration or you don't use that email address anymore. And that's the unfortunate thing is that we don't have, it's not perfect, and that's why we can't let go of these things. But yeah, I need to go through my Dropbox and delete about a terabyte of files right after this call because I have so many, I have so many. Like I, I have a Dropbox folder of just stock photos that I downloaded years ago that I haven't used yet, but I might use someday. It's like I don't even use stock photos anymore. And now I do everything AI generated and it's just, you forget, and you, especially when it's not a huge bill, like my S3 bills, like that's where you really gotta pay attention.'cause then you, I was like, why am I paying for this many gigs of data? What do I have on there? And I was like, oh my gosh, I had the same. Video file like six times of each video file and they're all gigabytes and big.'cause it's shooting 4K. So it all adds up. So I think this is, yeah, so the question is, we've now found the problem. How do we is it education, do we create, how do we create a culture where people are more conservative with their data usage or think about a process of storing, 'cause I wish, a great solution would be just if computers were like, you already have six copies of this file. Do you need six? But I've had a situation where I've tried to use those tools that really delete the duplicates and they delete both. So I don't trust those tools 'cause I've tried those, but it's that. So what do you think is the right path forward? I. I think education is a really big part of it, and there's some great organizations out there who are working on this exact problem. It was funny that we're talking about this as Cop 29, the big conference on climate is going on at the moment. Organizations like climate Tech Partners. I work with sustainable it.org and their responsible AI section, and there's a bunch of work we need we're putting together in terms of standards and PRI principles around sustainability. The Green Software Foundation, which is part of the Linux Foundation, is focused in on. Best practices for engineers to build green software and inquiry. And Sarah and Sarah I've forgotten. You, I don't have their names in front of me right now. They built a, they published a great book this year called Building Green Software that has some best practices. I also work with ISIC Global, which is an organization, a global one that certifies technology architects. We're, our aim is to make architecture of these large systems up at the scale of skill that you would do with a doctor or a lawyer that we actually have to go through a significant. Education process, and we've just launched the Sustainable Architects architectures group sustainable architectures.org is out there. Or if people are interested to set up the training and the best practices around what architects need to do to build much more sustainable systems. Out there. Yeah. Bunch of different organizations, all of them focused in on how do we use tech more responsibly and much more efficiently in it. I think this is, it's getting me a lot to think about because we just, especially in the West, like to US, storage and electricity are so cheap that we don't think about how everything is connected and that you don't have to lose quality of the ai, you don't have to lose the experience bad in conservatism, you can get just as great in experience by just using the right tools and having efficiency and not, and what slows down your computer when a hard drive is full. I'm always, that's when I start deleting stuff. I'm like, why is everything running so slow? Oh, I downloaded way too much stuff, too many games. So I, like my son, learned how to download Minecraft mods. So when I was using the PC computer the other day, I was like, what is, why is the hard, why is the main hard drive full? Oh my goodness. He's downloaded over a hundred gigabytes of Minecraft mods to the SSD, not to the data storage drive. Because he, to him, there's no difference and it's. Really small things can lead to really big things. So I think this is really helpful. I think this has given people some food for thought, which is important to see that a lot of times we think about conservatives and we think about party poopers and oh, the fun is all over, but it doesn't. Have to be that like it can be, there's still ways to get all the AI experience to just use the right models, to use the right tools so that we don't have any loss in experience. And honestly, like I just never wanna lose my files. If they're just organized better, I wouldn't have duplicates. And we just removed every duplicate from the internet and every like cloud server, we would probably save 90% of the space. So it's not even losing your data, just living, losing multiple copies and copies of copies. So I think this is really some good ideas to think about. Where can people find more about what you're doing, connect with you online and see some of the other projects you're doing as part of sustainable IT and some of the other majoring projects you're working on. I am a fractional CTO chief Technology Officer with Chief Li and Co, where we're all CXOs, fractional CXOs from various parts. And our focus is not only on sustainability, but profitability. How do we help organizations of all different sizes to. Lower their carbon footprint, use technology, use, product development, be aware of, their scope one, scope two, scope three greenhouse gas emissions calculations out there, and be able to make improvements over time. So that's what I do professionally. And by the way, we're when founded women run and we're all women CXOs. Because we're all really committed to sustainability going forward. You can also find me on sustainable architectures.org, which is part of the ISA Global, and helping raise the awareness of sustainability responsibilities by technology architects worldwide. And you can find me on LinkedIn Casey Clark. Amazing. Put links below everything below the video and in the show notes for today's episode. Thank you so much for being here for another amazing episode of the Artificial Intelligence Podcast. Thanks for listening to today's episode starting with ai. It can be Scary. ChatGPT Profits is not only a bestseller, but also the Missing Instruction Manual to make Mastering Chat, GBTA Breeze bypass the hard stuff and get straight to success with chat g profits. As always, I would love for you to support the show by paying full price on Amazon, but you can get it absolutely free for a limited time@artificialintelligencepod.com slash gift. Thank you for listening to this week's episode of the Artificial Intelligence Podcast. Make sure to subscribe so you never miss another episode. We'll be back next Monday with more tips and tactics on how to leverage AI to escape that rat race. Head over to artificial intelligence pod.com now to see past episodes. Leave, review and check out all of our socials.