Full Transcript for Can AI Scale Without Blowing Up Its Carbon Footprint?

00:02:02.778 --> 00:02:05.209 We, we are in a rush to adopt things. 00:02:05.239 --> 00:02:06.978 We adopt them all over the place. 00:02:07.028 --> 00:02:11.878 And then later we go back and decide what's useful, what's efficient, and then we kind of do that work. 00:02:11.979 --> 00:02:24.368 And so one of the common themes of all these trends has been this, the, the, the-waste we generate, the e-waste, the, the electricity overconsumption some of the impact we have on people and projects and processes along the way. 00:02:24.919 --> 00:02:34.049 And so it just seemed like, Hey, why don't we actually found some, some effort around that, the real issue of how do we get more effectively use the resources of the planet? 00:02:34.139 --> 00:02:37.609 How do we more effectively use the resources of our organization? 00:02:37.758 --> 00:02:41.269 And how do we ensure the work we're doing is benefiting you know, kind of the people and things around us? 00:02:41.748 --> 00:02:46.098 And and we were sitting around with a bunch of CIOs and, and, and asking about common problems. 00:02:46.098 --> 00:02:49.658 And one of the common problems was or common thoughts and problems they had was this. 00:02:50.644 --> 00:03:02.963 And then, and then lastly, maybe, maybe just you know, I'm of a certain age as are a number of my peers are kind of looking to the next generation and thinking, you know, maybe we didn't do such a great job as our job as stewards. 00:03:03.104 --> 00:03:10.649 So maybe we need to do a better job of stewarding resources for the next gen when that does kind of the background of how sustainable IT came, came about. 00:03:11.508 --> 00:03:17.729 And for people who are unaware, can you tell us what Sustainable IT is and does? 00:03:18.769 --> 00:03:26.308 Yeah, so, uh SustainableIT.org is a non-profit. 00:03:26.959 --> 00:03:30.618 We're created by, led by technology executives. 00:03:30.618 --> 00:03:36.521 And the idea is to advance sustainability around the world through technology leadership. 00:03:36.521 --> 00:03:39.609 So, lots of people are focussed on carbon, or decarbonisation, or yYou know, equity around the world, which is wonderful. 00:03:39.658 --> 00:03:44.334 But technology leaders often play a unique role in these sort of things. 00:03:44.334 --> 00:03:48.093 We, in many companies, you're the largest consumer of power. 00:03:48.324 --> 00:03:50.354 You own the e-waste stream and some other things. 00:03:50.394 --> 00:03:52.063 We didn't feel like we were doing enough. 00:03:52.313 --> 00:03:58.424 And so, you know, one of our, our kind of tenants is for too long, sustainability has been someone else's problem. 00:03:58.723 --> 00:04:01.183 We think it should be our problem and we should lean in. 00:04:01.183 --> 00:04:05.938 So we're creating specifically best practices, frameworks. 00:04:05.968 --> 00:04:10.449 We've established a standard and we've got a version of our standard published that you can consume. 00:04:11.598

--> 00:04:14.359 We're pursuing and creating certifications in this space. 00:04:14.699 --> 00:04:17.298 We're doing a lot of work on education, training, and awareness. 00:04:18.009 --> 00:04:22.218 We've engaged with the World Economic Forum, for instance, and attended Davos. 00:04:22.269 --> 00:04:25.639 We'll be back at Davos again this year, kind of pushing this message. 00:04:26.619 --> 00:04:39.928 And what's even more important in the era of AI, which you know, it seems to be an era these days AI is magical technology in some ways, but it's also incredibly energy intensive carbon intensive and e-waste intensive technology. 00:04:40.358 --> 00:04:45.999 And more and more we're being pulled into conversations and say, Hey, you guys, you're at the heart of the thing. 00:04:46.019 --> 00:04:48.639 We, we want to do a better job in a green economy. 00:04:49.199 --> 00:04:51.959 AI is wonderful, but like, it seems to be moving in the wrong direction. 00:04:52.369 --> 00:04:53.184 Can you guys help? 00:04:53.244 --> 00:04:55.990 So we're, we're having a lot of conversations about ethical AI. 00:04:56.321 --> 00:04:59.990 We've actually just released a white paper and are coming out with a conversation about sustainable AI. 00:05:01.706 --> 00:05:06.615 Okay, so seeing as you brought it up, what are you calling sustainable AI? 00:05:08.201 --> 00:05:08.500 Yeah. 00:05:08.500 --> 00:05:12.560 So, I guess AI is, we're early days. 00:05:13.680 --> 00:05:15.550 People have seen it do magical, interesting things. 00:05:15.550 --> 00:05:17.511 I can go create deep fakes. 00:05:17.511 --> 00:05:19.571 I can create, you know, great content. 00:05:19.951 --> 00:05:21.511 It's got some issues as well. 00:05:21.930 --> 00:05:31.704 So we've come out with essentially a framework and the framework, you know, trying to be catchy a bit and, and, memorable is reflect, reframe and reimagine. 00:05:32.413 --> 00:05:35.233 And the idea is, so reflect is pretty simple. 00:05:35.533 --> 00:05:39.314 Consider the intended uses and desired outcomes of your applications. 00:05:40.004 --> 00:05:49.189 It takes somewhere between 10 and 50 X, the amount of power and resource consumption to do a AI enabled search versus a non AI search. 00:05:49.718 --> 00:05:52.199 Does every search you do really need to be AI powered? 00:05:52.588 --> 00:05:58.579 Do you really need to use 50 times the amount of energy and therefore carbon or you know, or not. 00:05:58.858 --> 00:06:10.819 Reframe so, governance is a bit different in the world of AI because I'm not setting you know, because things are happening to some degree outside of my control, perhaps outside of my understanding in some of these models. 00:06:10.829 --> 00:06:11.139 So. 00:06:12.408 --> 00:06:21.278 Re-architecting rules, processes, roles and skill sets you know, thinking differently about architecture to maximize the benefits and minimize some of the negatives. 00:06:21.639 --> 00:06:25.468 And we've heard some of the negatives, you know, hallucinations and bias and things like that. 00:06:25.798 --> 00:06:32.742 And then reimagine conceptualize new business applications, processes and experiences that are AI, unique to AI's capabilities. 00:06:32.851 --> 00:06:41.127 So typically when we adopt early technology, the first thing we do is take our existing way of working and then skin it with the new technology. 00:06:41.317 --> 00:06:48.757 So an example of this might be if your Airbnb come comes out and you're a traditional hotelier, you're going to roll out your mobile app. 00:06:48.826 --> 00:06:49.487 What does it do? 00:06:49.487 --> 00:06:56.086 Well, mostly it just scrapes your website and gives you static information on the mobile app, but it takes you five or 10 years. 00:06:56.086 --> 00:06:57.487 And eventually what you've done is rebuilt it. 00:06:57.487 --> 00:07:02.026 So now it's my door key and I can chat with the concierge and I can do lots of things with that mobile environment. 00:07:03.216 --> 00:07:07.937 So reimagining says you want to get to that, that new version faster. 00:07:08.351 --> 00:07:11.021 Don't just reskin your existing processes with AI. 00:07:11.572 --> 00:07:16.468 And, and in doing so, you may in fact be able to use AI for good, so to speak more quickly. 00:07:16.668 --> 00:07:20.149 And then under that we have nine commitments of deployment. 00:07:20.579 --> 00:07:35.245 Do due diligence on risk, do due diligence on sustainability, do due diligence on ethical usage optimize your data for AI so that you're being efficient there build trustworthy outcomes, drive AI literacy across your company and user base. 00:07:35.785 --> 00:07:41.985 And then be human first, be inclusive, and innovate responsibly. 00:07:42.014 --> 00:07:46.911 So, worry about the impact you're having on the organization and people before you do it. 00:07:47.987 --> 00:07:48.297 Okay. 00:07:48.297 --> 00:07:50.026 And will it be available online? 00:07:50.036 --> 00:07:54.987 I mean, can I include a link in the show notes, for example, for people who are interested to check it out? 00:07:55.982 --> 00:07:57.291 Yeah, absolutely, absolutely. 00:07:57.322 --> 00:07:59.482 Like I said, we've got a white paper, it should be up on the website. 00:07:59.482 --> 00:08:02.732 Actually, if it's not up today, it'll certainly be up by the time the podcast is out. 00:08:02.771 --> 00:08:03.521 You can get a link for that. 00:08:04.047 --> 00:08:04.336 Cool. 00:08:04.526 --> 00:08:04.867 Perfect. 00:08:04.867 --> 00:08:05.127 Great. 00:08:05.637 --> 00:08:13.466 And talk to me a little bit about standards, because I know that's something that, you know, you guys are quite bullish about. 00:08:13.937 --> 00:08:22.067 Do you see them as a, a necessary burden or as a potential game changer in how businesses approach sustainability? 00:08:23.086 --> 00:08:26.416 Yeah, I think it's an opportunity, right? 00:08:26.466 --> 00:08:29.307 Which is, which is a polite way

of saying both. 00:08:30.697 --> 00:08:37.063 You know, I think John Chambers once said education is what you get when When things don't turn out the way you expected. 00:08:37.203 --> 00:08:40.653 So, so, this is sort of fits that, that criteria a bit. 00:08:40.774 --> 00:08:43.313 So, so, one, it's an opportunity, right? 00:08:43.313 --> 00:08:53.793 We, we should be measuring what we're doing, be thoughtful about what we're doing, and unless you're gathering data and, and then organizing it in a way where you can compare it with others it's, it, it's difficult to understand if I made progress. 00:08:54.163 --> 00:08:54.183 Um, 00:08:54.774 --> 00:08:55.813 Back to the UN. 00:08:55.813 --> 00:09:03.553 The UN established the Paris Climate Accord some decades ago, a set of objectives around carbon reduction and so on. 00:09:04.053 --> 00:09:10.083 And then they did what they call a stock take, a check in to see how we're doing a few years back. 00:09:10.083 --> 00:09:13.004 So on the kind of 20th year anniversary, we went, okay, how are we doing? 00:09:13.543 --> 00:09:15.173 Not surprisingly, not great. 00:09:15.774 --> 00:09:19.114 So we made some progress, but nowhere near what was expected or what was committed. 00:09:19.839 --> 00:09:26.168 And part of the reason is, you know, we're not consistently measuring it in a way we're not holding each other accountable annually. 00:09:26.168 --> 00:09:30.254 And so now they're, they're driving to do some of those things, things more actively. 00:09:30.794 --> 00:09:39.044 So, so when we set down to set the standard, one of the things we found an old adage I like to quote the great thing about standards are there's so many to choose from. 00:09:39.445 --> 00:09:41.450 And, and the same, the same is true for reporting. 00:09:41.450 --> 00:09:45.379 Like, you know, you're reporting for, oh, I operate in the EU so I have to do this. 00:09:45.379 --> 00:09:47.389 The UK is different, Germany's different. 00:09:47.779 --> 00:09:53.570 The US, California is different than the rest of the US, you know, so there's a million standards and a million kind of things you've gotta address. 00:09:54.139 --> 00:09:55.370 So we looked to make it easy. 00:09:55.370 --> 00:10:02.070 We said, great we're gonna take some of these core standards, we're gonna extract from them the bits that are relative to this conversation. 00:10:02.070 --> 00:10:08.609 So, so I may have a 28 page reporting standard I've gotta do for my 10 k or for something else. 00:10:08.908 --> 00:10:12.816 And there's two sentences or two paragraphs on IT, or two paragraphs on carbon. 00:10:13.316 --> 00:10:14.716 So we pulled them all out into one place. 00:10:14.755 --> 00:10:17.365 We, we organized it in a way that we think makes sense. 00:10:17.365 --> 00:10:21.135 There's an environmental, a governance, a societal branch. 00:10:21.216 --> 00:10:25.525 And each of those branches then allow you to go kind of double click on those standards. 00:10:25.525 --> 00:10:34.125 And every one of the data elements we recommend you gather links out to each of the standards and where in those standards that you're required to do as well. 00:10:34.125 --> 00:10:44.870 So, our belief is if you follow our standard and you publish that data, you're going to be 80 or 90% compliant with everything else you need to report anyway, because that data's there and you'll be able to kind of cross link it. 00:10:45.431 --> 00:10:49.697 And then our commitment is as new things come out, we'll be revising and updating our standard as well. 00:10:50.376 --> 00:10:59.787 And the beauty is if you're an IT leader you already own the standard, because we've partnered with ServiceNow, and ServiceNow has adopted our standard as part of their core platform. 00:11:00.211 --> 00:11:07.052 And so you can literally go into, if you're using ServiceNow, you can go in and kind of flip a switch and it can start telling you kind of how you're doing in some of these key areas. 00:11:35.862 --> 00:11:36.272 Nice. 00:11:36.342 --> 00:11:37.002 Okay, cool. 00:11:37.672 --> 00:11:42.432 And what about the increasing pressure of regulations and reporting requirements? 00:11:42.802 --> 00:11:45.042 How are the CIOs that you speak with? 00:11:45.042 --> 00:11:47.362 How are they navigating these challenges? 00:11:48.472 --> 00:11:50.822 Yeah, it's a, it's a challenge. 00:11:50.932 --> 00:11:59.792 And I think that's one of the reasons people are coming to us and embracing this conversation is this sense of not only feel like we're going to catch up on the new standards all the time. 00:12:00.262 --> 00:12:01.772 It'd be great at some point to get in front of it. 00:12:02.352 --> 00:12:20.596 And we think it's an opportunity here to actually lead, to demonstrate to those standards bodies, to those regulatory groups, hey, we care, we do care about this conversation, and we are compliant, and we are doing this work, because most of the times they'll roll out a new standard, and there's a deadline, but then there's a grace period. 00:12:20.806 --> 00:12:25.176 You know, it's a deadline, but if you're showing intent toward working toward this, we're going to give you grace. 00:12:25.666 --> 00:12:38.236 So we think if you've got a, an active forward looking posture around gathering data, driving improvement if you don't tick a box on, you know, on a new standard perfectly, you'll probably get grace around, around evolving to that. 00:12:39.706 --> 00:12:40.036 Okay. 00:12:40.636 --> 00:12:41.066 Okay. 00:12:41.866 --> 00:12:44.646 You mentioned earlier the problem of e-waste. 00:12:45.076 --> 00:12:49.046 It's a, it's obviously a huge, huge problem, especially with the tech boom. 00:12:49.346 --> 00:12:51.686 And as you mentioned, AI's rapid growth as well. 00:12:52.266 --> 00:13:00.566 How can IT leaders turn this challenge into an opportunity, or can they to reshape consumption and disposal behaviors? 00:13:01.596 --> 00:13:07.292 Yeah,

the production of, and disposal of AI related hardware in particular is driving this, like, through the roof. 00:13:07.709 --> 00:13:12.559 It was bad enough before, right, with all our mobile phones and, you know, our consumption economy. 00:13:13.149 --> 00:13:22.139 The carbon emissions and e-waste the global volume of that today is predicted about 120 million tons annually. 00:13:22.949 --> 00:13:26.259 And by 2050, we think it'll be double what it is today. 00:13:26.979 --> 00:13:28.559 And 2050 isn't that far away. 00:13:29.089 --> 00:13:35.234 So the other really disappointing fact is less than 20 percent of that is formally recycled today. 00:13:36.274 --> 00:13:39.214 So big opportunity day one right now. 00:13:39.234 --> 00:13:40.934 You don't have to wait for anything technology. 00:13:40.934 --> 00:13:42.734 Is it like engage with a partner? 00:13:42.734 --> 00:13:55.164 We've, we've got a partnership with a company called Human AI, great organization globally that will partner with companies and help solve this some of the e-waste stream problem recycling, or more importantly, reuse problem. 00:13:55.204 --> 00:13:56.404 So, so these guys are wonderful. 00:13:57.074 --> 00:14:03.554 I tried as a CIO before to donate a number of out of cycle laptops to a local charity. 00:14:03.554 --> 00:14:08.454 And after six months of trying to go through hoops, I finally gave up. 00:14:08.584 --> 00:14:09.264 Oh, wait a minute. 00:14:09.764 --> 00:14:12.024 You've got these under a corporate Microsoft license. 00:14:12.074 --> 00:14:13.824 They don't have an operating system with them. 00:14:14.114 --> 00:14:17.594 Oh, wait, you know, like the, the asset tag issue. 00:14:17.634 --> 00:14:22.094 Like, like, what if something, what if it ended up in a land landfill later and it comes back? 00:14:22.144 --> 00:14:24.264 Like it was just impossible. 00:14:24.314 --> 00:14:25.584 And these guys have solved that problem. 00:14:26.189 --> 00:14:31.029 So they, they take, they strip the asset, they, they, they take ownership of the asset, so it's now owned by them. 00:14:31.539 --> 00:14:38.619 They then transfer ownership, they solve the operating system problem by some licensing agreements they have as a non profit. 00:14:38.959 --> 00:14:46.719 And they're doing great work taking, what to us is a three or four year old laptop might be considered no longer enterprise grade. 00:14:47.109 --> 00:14:49.249 It's got another four or five years of life on it. 00:14:49.369 --> 00:14:53.954 If you're a, if you're somebody who doesn't have access to the internet because you don't have a computer at home. 00:14:54.584 --> 00:14:57.404 So, huge, huge opportunity there. 00:14:57.604 --> 00:15:04.864 Additionally people are starting to do things like, Hey, it's four years, like, computers these days are quite powerful. 00:15:05.214 --> 00:15:07.674 Do I really need another computer every four years? 00:15:08.014 --> 00:15:10.494 And do I need a whole new computer?

00:15:11.054 --> 00:15:18.044 So, you know, I'm very Apple, of course, has been out in front in this conversation about getting to, to, to carbon zero. 00:15:18.054 --> 00:15:23.049 So, the iPhone 15 Pro is one of the most repairable phones ever made. 00:15:23.529 --> 00:15:26.749 So it used to be you cracked the screen, in fact, virtually had to throw your phone away and get a new one. 00:15:27.399 --> 00:15:34.539 Now it's designed and engineered so almost any problem the phone has can be repaired in under two hours at the at the Apple store. 00:15:35.319 --> 00:15:44.474 So, that's the sort of thing that will get us to where we're going or at least taper off the growth of what's happening is if we can start looking at, hey, how can I extend the life of these? 00:15:44.794 --> 00:16:03.701 So, for instance, today, people that are buying, you know, that are queuing up and haranguing NVIDIA so they can get their allocation of GPUs to go build their AI farm should be thinking today about what happens in three years when those GPUs are no longer cool, and I want to buy the cooler GPU, where do those GPUs go? 00:16:03.972 --> 00:16:05.402 Like, is there a secondary use case? 00:16:05.442 --> 00:16:08.122 And where should I be thinking about those versus scrapping those? 00:16:08.342 --> 00:16:09.642 And putting them in a landfill somewhere. 00:16:10.617 --> 00:16:12.117 Yeah, yeah, no, absolutely. 00:16:12.117 --> 00:16:12.707 Yeah. 00:16:12.887 --> 00:16:13.247 Yeah. 00:16:13.326 --> 00:16:14.647 On, on the phone issue. 00:16:15.362 --> 00:16:24.422 It's the, the right to repair is obviously a huge one, but also they're starting to be designed as well for take apart at end of life, aren't they? 00:16:25.062 --> 00:16:25.382 Yeah. 00:16:25.701 --> 00:16:26.002 Yeah. 00:16:26.342 --> 00:16:34.442 Yeah, because one of the reasons that so few things are, are recycled is the, the cost essentially to disassemble and sort the bits. 00:16:34.822 --> 00:16:45.182 And so because because that's a big issue yeah, they're designed now so they're a lot easier to sort into the recyclable component bits so that they can more, more easily be done in an economically viable way. 00:16:46.057 --> 00:16:47.787 Yeah, Yeah, yeah, yeah. 00:16:47.817 --> 00:16:49.957 It's it's great to see advances there. 00:16:50.447 --> 00:16:55.377 What about, I mean, you talked about AI and it is a hot topic right now. 00:16:55.757 --> 00:17:12.316 How should CIOs balance the promise of AI with the the need for sustainable practice, I mean, particularly as, as you said, already with AI, causing so much e-waste and also it's massive power consumption. 00:17:13.266 --> 00:17:23.959 Yeah, yeah so, so again, some of the data essentially since 2012 the AI training runs are exponentially consuming more power. 00:17:24.243 --> 00:17:27.874 So they're doubling every three and a half, about 3.4 months. 00:17:27.874 --> 00:17:29.223 I mean, just crazy curve. 00:17:29.584 --> 00:17:31.859 Even

GPT three used about 1. 00:17:31.859 --> 00:17:34.269 3 gigawatt hours to train. 00:17:34.769 --> 00:17:37.588 GTP4 is 10X larger than that. 00:17:37.879 --> 00:17:46.459 So GTP4 used just, you know, the same amount of power as a small city for about four months of a small city's power, just to train that model. 00:17:46.959 --> 00:17:56.939 Google AI today burns around 2.3 terawatts of power annually which is as much, as much electricity as it takes to power the city of Atlanta for, for a year. 00:17:57.399 --> 00:18:01.139 So, it's just kind of like getting your head around how big these numbers are. 00:18:01.139 --> 00:18:01.769 It's huge. 00:18:02.558 --> 00:18:03.479 What can you do about that, right? 00:18:03.499 --> 00:18:04.879 So, so one we've talked about already. 00:18:04.909 --> 00:18:06.709 Hey, AI doesn't fit every problem. 00:18:06.709 --> 00:18:08.798 Let's not do, let's not do, do that. 00:18:08.849 --> 00:18:20.078 There are a bunch of things researchers are doing to improve on that, but both because, you know, it's good for the world and good for the economy, but also, frankly, it's good for their bottom line because it costs literally billions of dollars to train these models. 00:18:20.509 --> 00:18:23.599 Hey, what if I could figure out a way to do it more efficiently? 00:18:23.939 --> 00:18:33.389 And so, so one of the things today is, you know, these models on a very oversimplified level are taking, you know, one word in a document, kind of correlating it mathematically to every other word in the document and going to word number two and doing the same. 00:18:33.519 --> 00:18:42.019 Well, what if I knew three steps in which branch was going to be a dead end and I could prune that branch and not, not, not train that branch. 00:18:42.349 --> 00:18:51.548 So using techniques like that, you can save, literally you can be 80% more efficient to power consumption and get to about a 90 to 95% as a good model. 00:18:52.084 --> 00:18:53.413 And maybe that's good enough. 00:18:53.413 --> 00:18:57.754 And certainly considering the money and the power savings it's certainly worth considering. 00:18:57.774 --> 00:18:58.683 People are looking at that. 00:18:59.144 --> 00:19:03.433 There are also solid state models where they're training very differently that are much, much less energy. 00:19:03.504 --> 00:19:05.884 So we're getting better at understanding that piece. 00:19:05.953 --> 00:19:16.943 The analogy I would use, and I think the lesson for most CIOs is or IT leaders is it's like the first time you use the cloud all of a sudden there's this like virtually free container or VM. 00:19:17.423 --> 00:19:20.074 And really cheap storage sitting out in the cloud. 00:19:20.433 --> 00:19:25.864 And so I could go consume that and it just seemed like, ah, these, it's virtually free resources and they're infinite. 00:19:26.224 --> 00:19:26.963 This is amazing. 00:19:27.784 --> 00:19:33.054 And then one day, maybe a month later, a quarter later, two quarters later, you woke up and got a bill from Amazon. 00:19:33.453 --> 00:19:35.314 Your, your first million dollar AWS bill. 00:19:35.314 --> 00:19:37.423 And you went, what the, where did, oh my goodness. 00:19:37.884 --> 00:19:42.663 And, and so then you start applying governance and you start turning off these things that aren't being used and so on. 00:19:43.469 --> 00:19:44.919 So, same applies here. 00:19:45.499 --> 00:19:48.068 You turn on AI, you encourage everybody in the company to start using it. 00:19:48.409 --> 00:19:51.939 And, and I'm getting almost free, my API calls are almost free. 00:19:52.409 --> 00:19:54.808 And, and almost is the keyword, right? 00:19:54.808 --> 00:20:00.449 And then one day you go, wait a minute, okay, now we're going to restrict AI to only the legal team and finance because you guys are... 00:20:01.048 --> 00:20:02.828 so, so, think about that up front. 00:20:02.879 --> 00:20:07.068 Like, let's govern usage, let's put in smart, thoughtful processes. 00:20:07.068 --> 00:20:09.338 Yes, you want to encourage use, but do it in a thoughtful way. 00:20:10.229 --> 00:20:21.869 Okay, do you see quantum computers stepping in at some point and taking over AI because they would be, I would have assumed far more efficient at doing some of those processes? 00:20:22.389 --> 00:20:23.479 Yes and no. 00:20:23.639 --> 00:20:28.298 So quantum computers won't be at the scale for some time. 00:20:28.298 --> 00:20:33.229 So, so quantum computers are very good at certain things, certain mathematical difficult mathematical problems. 00:20:33.749 --> 00:20:40.209 But they don't have the massive parallelism that, you know, that GPU, I mean, GPUs win because of the, the sheer number of cores. 00:20:40.239 --> 00:20:42.139 Just, you know, a convincing number of cores. 00:20:42.642 --> 00:20:48.292 The real challenge and opportunity with quantum computing, as I was saying, it will do some things very differently. 00:20:48.332 --> 00:20:53.402 So it'll be good at some things and really poor at other things that current computers are inversely skilled at. 00:20:53.852 --> 00:21:00.919 The real thing that people are starting to worry about is cyber security as it relates to quantum. 00:21:00.979 --> 00:21:07.288 And I've started to see people requesting in RFPs and so on, requesting things to be quantum safe. 00:21:07.899 --> 00:21:21.794 Meaning in the inevitability that quantum computers become more available, today's 256 bit, you know, encryption keys are assuming current computers and it takes, you know, years to, to brute force hack it. 00:21:21.814 --> 00:21:23.804 Well, under quantum computing, maybe it's hours. 00:21:24.394 --> 00:21:25.334 So, now what do I do? 00:21:25.884 --> 00:21:32.729 And I think, I, I don't know that we have terribly good answers to that just yet, but a lot of folks, I think, are worrying about it and, and I think for good, for good reason. 00:21:32.859 --> 00:21:36.939 We, we know there

are state agencies that have literally archived. 00:21:37.844 --> 00:21:44.676 Terabytes of encrypted data just waiting for the opportunity to decrypt it once the technology is available. 00:21:44.676 --> 00:21:47.950 Yeah, Everyone Ralph loves a good story. 00:21:48.580 --> 00:21:59.285 Do you have a favorite example of a CIO member, for example, how their sustainability efforts might have led to a meaningful outcome, you know, and what could others learn from that success? 00:21:59.285 --> 00:22:05.362 Yeah, a couple come to mind . So one is maybe not the outcome, but the, the driver. 00:22:05.372 --> 00:22:09.644 One of the things people struggle with is, how do I get momentum behind an initiative? 00:22:09.644 --> 00:22:11.904 Like I care, but how do I get everybody else in my company to care? 00:22:12.455 --> 00:22:12.904 And. 00:22:13.845 --> 00:22:22.599 I was in Germany and in fact at a meeting in Germany in Jan-Feb last year and spoke with the CIO of the German rail system. 00:22:23.009 --> 00:22:25.225 And she was saying they countrywide. 00:22:25.225 --> 00:22:33.832 Germany has an initiative to reduce auto transport by some, I forget the number, some staggering number, like 30 or 40 percent over a very short amount of time, three or four years. 00:22:33.971 --> 00:22:41.207 That's impressive that as a country, you guys are, you know, rallying behind the cause of sustainability so, so vehemently. 00:22:41.667 --> 00:22:44.767 And she went, Oh, no, like, yeah, we care, but that's not why. 00:22:45.336 --> 00:22:46.737 And I said, Well, then why? 00:22:46.937 --> 00:22:57.576 Well, because, you know, the war in Ukraine means we no longer have access to you know, oil to create petrol, and our whole economy and way of life is at risk if, you know, we don't do something about this. 00:22:57.596 --> 00:22:59.947 And I went, Okay, now there's motivation, right? 00:23:00.207 --> 00:23:02.547 It's almost a live or die kind of proposition. 00:23:02.547 --> 00:23:09.912 And I think so obviously not everybody has that kind of lever but it does remind you that never waste a good crisis, you know, kind of advice. 00:23:09.942 --> 00:23:28.261 If something is driving change in your industry, some competitive threat is out there, some like find a way to connect some of what you're doing to those opportunities and you'll get a lot more traction of, of going but, but probably the best example, and this isn't a typical example, there's a company we were talking to. 00:23:28.701 --> 00:23:31.162 And they're not public about this, this, what they've done. 00:23:31.162 --> 00:23:49.451 So I, I can't mention names, but they were running away for fat in, in Asia and they were trying at a green initiative and they really wanted to use more you know, more, more green power and they, there just wasn't, it's not available in the grid because everybody else wants to buy green power and there's not enough solar or wind in the local, in the local grid. 00:23:50.201

--> 00:23:56.572 And so they're kind of scratching their head and, but by now they've talked to the, all the generators, they've talked to the. 00:23:57.197 --> 00:24:06.667 The power grid company, they've talked to their own floor and they started and they gathered a bunch of data and they decided almost on a whim. 00:24:06.677 --> 00:24:09.027 They said, well, what if we let's go ask AI this problem? 00:24:09.037 --> 00:24:10.427 AI is supposed to be so smart. 00:24:10.707 --> 00:24:15.336 And they looked at a bunch of data that they gathered into their AI model and then trained it and did some work. 00:24:15.557 --> 00:24:19.927 And ultimately they came back with a, an idea that they hadn't thought of. 00:24:19.997 --> 00:24:30.312 And the idea was, well, what if instead of just trying to buy, you know, reduce power or buy green power, what if we could essentially live in the valleys of, of of demand? 00:24:30.772 --> 00:24:36.142 And by that, you know, the typical power grid, there's peaks of demand and then there's, you know, valleys of usage. 00:24:36.461 --> 00:24:46.632 And the notable thing about the power grid, certainly in most develop, developing and developed countries, is there's no buffer capacity. 00:24:47.201 --> 00:24:49.211 So when you turn on your light switch. 00:24:49.751 --> 00:25:00.672 It's not like you're drawing power that was stored some time ago and until and while somebody else, you're, you're, you turn that on power is essentially coming out of the reactor over the lines to your house. 00:25:00.711 --> 00:25:03.062 Like there's no buffer capacity in the middle to speak up. 00:25:03.892 --> 00:25:12.522 And so, so what they were able to do using AI is look at the the data, the use of power from every machine and device on the factory floor. 00:25:13.042 --> 00:25:34.282 Match that against the power distribution grid when it was busiest or in least demand and match it against the same with the power generators and they were able to, at the same production output level, the same amount of output, were able to reduce essentially energy consumption by almost 80%. 00:25:34.312 --> 00:25:39.142 Now they're using the same amount of power but they're using it at a lower cost of power. 00:25:39.711 --> 00:25:42.942 Because they're using it in these troughs where, where it's demand pricing. 00:25:43.461 --> 00:25:47.991 And, and, and that, it meant they didn't have to bring another reactor online. 00:25:47.991 --> 00:25:52.652 They don't have to build another, you know, coal fired power plant et cetera, in that region. 00:25:53.052 --> 00:25:55.822 Because they're more efficiently using the power when it's available. 00:25:56.052 --> 00:26:06.442 Which is, so again, not everything is about net reduction of, of carbon footprint or net, like, sometimes it's just, hey, let's optimize for the resource use that's available to us. 00:26:07.037 --> 00:26:08.947 to prevent the need to go create more. 00:26:09.567 --> 00:26:25.757 Yeah, yeah, no, I think particularly when talking about electricity grids, I think the demand side management is an underutilized possibility resource, whatever you want to call it, because it's not being done enough. 00:26:26.147 --> 00:26:29.647 But I think going forward, it's going to be something that's going to be very, very powerful. 00:26:31.017 --> 00:26:38.126 There's a former Tesla one of Tesla's early battery scientists and engineers left Tesla and formed his own company. 00:26:38.576 --> 00:26:43.636 And what they're trying to create is a essentially economical battery platform. 00:26:44.027 --> 00:26:47.917 So there's more battery storage distributed around neighborhoods. 00:26:48.366 --> 00:26:50.096 And it's using a very different kind of batteries. 00:26:50.136 --> 00:26:53.717 It's not the nickel, it's not the The lithium and kinda rear metal kind of stuff. 00:26:53.717 --> 00:26:55.037 It's using literally rust. 00:26:55.037 --> 00:27:08.427 It's a, it's using iron oxide to create these much cheaper easier to store kind of batteries that have longer life and are kind of environmentally more friendly, but may help help solve some of this grid you know, this kind of grid imbalance problem. 00:27:09.106 --> 00:27:10.967 Yeah, iron air batteries, aren't they? 00:27:11.817 --> 00:27:12.247 Yeah. 00:27:12.247 --> 00:27:12.407 Yeah. 00:27:13.707 --> 00:27:14.177 Heard about them. 00:27:14.267 --> 00:27:14.596 Very good. 00:27:15.537 --> 00:27:30.280 So for your CIOs or for organizations who are looking at this, what kind of metrics or KPIs should they focus on to measure the success of their sustainability initiatives and that, you know, how can they ensure continuous improvement? 00:27:31.615 --> 00:27:31.895 Yeah. 00:27:31.895 --> 00:27:35.395 So, you know, again, again, somewhat self serving. 00:27:35.405 --> 00:27:36.796 I would put you back to our standards. 00:27:37.186 --> 00:27:40.135 I think there's a bunch of them there. 00:27:40.436 --> 00:27:41.706 And often we'll get the question. 00:27:41.706 --> 00:27:42.436 Where do I start? 00:27:42.666 --> 00:27:43.946 You kind of start with what you have. 00:27:44.076 --> 00:27:44.756 Turn on what you have. 00:27:44.756 --> 00:27:45.296 Start there. 00:27:45.750 --> 00:27:47.971 And then look at the areas that where you have gaps. 00:27:47.980 --> 00:27:52.780 So there'll be a lot of data that you don't have access to, or you have to go create manually, which is fine. 00:27:52.831 --> 00:28:09.551 Sometimes it's worth doing that, that seed run to go create a bunch of manual data to see where you are on your carbon uses, your e-waste or or, and some, some industries water usage is a big, a big issue and increasingly an issue in, in the AI conversation because. 00:28:10.346 --> 00:28:15.115 It used to be located your data center near the fiber optic lines because you wanted low latency to the internet. 00:28:15.395 --> 00:28:18.046 And now you locate your data center where there's

power. 00:28:18.326 --> 00:28:23.336 Because you need to power these big, big, big AI farms and AI farms. 00:28:23.666 --> 00:28:29.816 Basically, one of the problems in computing is power translates to heat because these things dissipate a lot of heat. 00:28:30.706 --> 00:28:31.516 And now we got to cool that. 00:28:31.516 --> 00:28:33.610 So now I'm running water through to cool it. 00:28:33.921 --> 00:28:42.441 And in some rural areas, all of a sudden the little data center that got built, you know, in the prairie is consuming 10% of the water, of the, the local municipality. 00:28:42.530 --> 00:28:43.611 So, big issue. 00:28:44.221 --> 00:28:45.901 So, so look at what matters to you. 00:28:46.141 --> 00:28:49.441 Begin measuring that, and if you're doing it manually, fine. 00:28:49.561 --> 00:28:50.881 But then find a way to automate that. 00:28:50.941 --> 00:28:54.961 Find a way to begin accessing that data in a, in a, in a more automated way. 00:28:55.301 --> 00:28:57.942 Because over time that's what really helps is this, it's the, the. 00:28:58.556 --> 00:29:03.206 Looking at the chart and looking in the trend and understanding what you can do about making the needle move in the right direction. 00:29:04.280 --> 00:29:05.020 Okay, cool. 00:29:05.230 --> 00:29:05.520 Nice. 00:29:07.000 --> 00:29:08.421 Quick left field question. 00:29:09.311 --> 00:29:19.230 If you could have any celebrity or fictional character as a spokesperson for your project, who would it be and why? 00:29:21.476 --> 00:29:22.395 That's a good question. 00:29:22.716 --> 00:29:23.976 It is bit left field. 00:29:24.026 --> 00:29:32.846 You know, there are so many kind of celebrity spokespeople that I think, you know, in certain topics, people, people get aren't particularly amenable to that. 00:29:33.276 --> 00:29:50.875 If I look back historically what I would say is somebody politically, if I go back to like, like US president Roosevelt, who looked at a big problem you know, the global recession depression, Great Depression, and didn't try to solve it in a kind of haphazard way. 00:29:50.875 --> 00:29:55.226 He thought, really, you know, thought it through and said, Hey, we're going to create a social safety net. 00:29:55.226 --> 00:29:58.855 We're also going to put people to work, to do infrastructure projects. 00:29:58.855 --> 00:30:00.506 We're going to pay them for that work. 00:30:00.506 --> 00:30:04.566 They're going to have pride and ownership of what they've built, they didn't get a handout. 00:30:04.709 --> 00:30:08.798 So he thought about the problem in this sort of circular way. 00:30:09.379 --> 00:30:13.399 He looked kind of at the circular economy and said, how do I solve all the problems we're having? 00:30:13.499 --> 00:30:16.308 Not just the surface level of the problem. 00:30:16.308 --> 00:30:24.349 And that's really what we're trying to do is like, look at the holistic problem and provide holistic solutions, not, not just easy solutions kind of off the top. 00:30:25.149 --> 00:30:27.318 Nice, good one, yup I like it. 00:30:27.919 --> 00:30:30.054 We're coming towards the end of the podcast now, Ralph. 00:30:30.074 --> 00:30:38.594 Is there any question I did not ask that you wish I had or any aspect of this we haven't covered that you think it's important to make people aware of? 00:30:38.594 --> 00:30:40.209 Yeah. 00:30:40.374 --> 00:30:47.844 the one thing that we have, the area we haven't gotten into which I'd used to be quite happy to avoid is the area these days of politics, right? 00:30:47.844 --> 00:30:53.663 And geopolitics is unfortunately becoming a bigger part of our lives because behaviors are driving a lot of things. 00:30:54.304 --> 00:31:00.274 And one of the things that's happening in the US and increasingly spreading to the EU is this sort of ESG backlash. 00:31:00.514 --> 00:31:06.193 In the US, there's this backlash against what they call the woke agenda and a bunch of other things that we're getting lumped into. 00:31:07.213 --> 00:31:13.153 And so, so, you know, what should we do about that, I guess, would be the question I would ask. 00:31:13.153 --> 00:31:18.264 And I think how I would answer that is it don't, don't kind of, don't let people paint us into a corner. 00:31:18.264 --> 00:31:20.134 Don't let people tell us what we stand for. 00:31:20.134 --> 00:31:26.394 We, we aren't you know, just out there, do gooders out there, like, trying to save the world kind of thing. 00:31:26.443 --> 00:31:31.504 Most of us are, you know, we're practical people that have businesses to run or businesses we contribute to. 00:31:31.814 --> 00:31:43.463 And what we're doing is we're doing what we're doing we think is smart and, and ethical and reasonable for not just the short term profit and benefit of the company and the world, but also the long term profit and benefit. 00:31:43.474 --> 00:31:53.453 So, don't let people frame you in a corner, like reframe the conversation around the other reason we're doing, you know, we worry about carbon footprint is among other things, I can be a more efficient at consuming. 00:31:54.713 --> 00:31:55.294 power. 00:31:55.544 --> 00:32:09.324 I can also, if I operate in a world that's increasingly hotter and increasingly more polluted, what does that do from the long term health and benefit of my company, employees, customers, et cetera. 00:32:09.334 --> 00:32:15.084 So, like, we're just thinking we have a longer term lens than you do if you're not worried about those sort of things. 00:32:15.134 --> 00:32:26.314 And so reframe the conversation away from politics and more toward Hey, we're doing this because like we're trying to be thoughtful, smart people about the work we're doing, and we're trying to create positive outcomes for our customers and communities. 00:32:27.949 --> 00:32:36.199 Super Ralph, if people would like to know more about yourself or any of the topics we discussed on the podcast today, where would you have me direct them? 00:32:37.294 --> 00:32:39.284 Yeah, first of all, sustainableit. 00:32:39.294 --> 00:32:39.693 org. 00:32:39.774 --> 00:32:46.693 You can read up on much of what we've talked about there and, and stay connected for future updates as we release more more information and so on. 00:32:47.183 --> 00:32:48.984 I can be found fairly easily on LinkedIn. 00:32:49.014 --> 00:32:51.244 It's just Ralph Loura, L O U R A. 00:32:51.744 --> 00:32:55.084 And I'm happy to connect there and and see if we can further each other's missions. 00:32:55.899 --> 00:32:56.328 Tremendous. 00:32:56.749 --> 00:32:57.798 Ralph, that's been fantastic. 00:32:57.849 --> 00:32:59.269 Thanks a million for coming on the podcast today. 00:33:00.744 --> 00:33:01.584 Thanks Tom, I really enjoyed it. 00:33:02.593 --> 00:33:04.462 Okay, we've come to the end of the show. 00:33:05.182 --> 00:33:06.293 Thanks everyone for listening. 00:33:07.153 --> 00:33:15.222 If you'd like to know more about the Climate Confident podcast, feel free to drop me an email to tomraftery at outlook. 00:33:15.472 --> 00:33:18.952 com or message me on LinkedIn or Twitter. 00:33:19.523 --> 00:33:27.763 If you like the show, please don't forget to click follow on it in your podcast application of choice to get new episodes as soon as they're published. 00:33:28.393 --> 00:33:31.353 Also, please don't forget to rate and review the podcast. 00:33:31.553 --> 00:33:34.083 It really does help new people to find the show. 00:33:35.173 --> 00:33:35.623 Thanks. 00:33:36.293 --> 00:33:37.212 Catch you all next time.