AI: Art of the Possible. Delivery of the Practical

AI: Art of the Possible. Delivery of the Practical

00:00 Well, thank you, Allice, and I want to thank all of you for having me here today. I have been as an analyst, coming to Canada and interacting with CIOs and IT leaders for over 35 years, and every time, I’ve always come away from my meetings and sessions with a better understanding of the future of the IT industry, and also, I believe, a better feeling about the future of the IT industry. And I have no idea, no doubt that’s going to be the same today. So I’m looking forward to a lot of great conversations and chats. My role is to really kick off with the topic of the day, which is AI. And the question we all are going to be asking here in a second, let’s make sure that’s rolling, is, is AI in the future going to be a conversation where we talk about how it has become a critical part of what we all do with technology and how we live our lives? Or is it something where we never quite got the promise and was delivered on what we wanted to have here? So what we’re looking for is to really spend a little time talking about that, but to set the stage, let’s look back now 18 months ago and ask the question, what were we all doing when this sort of whole focus on Gen AI came to the fore, and people really wanted to emphasize that. Well, what we did is there were a lot of conversations about how this was going to drive an amazing amount of GDP growth, and it was going to improve the productivity and we had announcements from people like McKinsey saying there’s this massive impact on the productivity of employees and staff, and we are doing all these use cases and test cases and things to make that work. And in reality, when we looked at it, what we did over the last 18 months was do a lot of use cases. Hundreds of use cases were analyzed and assessed, and then we actually went and did proof of concepts, not for all of those, but because we had limited budget, we’d go and focus on some specific ones. And what IDC did is, back in April of this year, we did a status check on that work. So we went out and we interviewed about 1,000 IT leaders around the world, in different regions, different countries, and asked them, how many proof of concepts have you done? How many conversions to production have you achieved by that point? And how satisfied are you with those production launches? So I’m showing you here what were the results in Canada. So about 34 on average, about 33 proof of concepts were launched that translated into about four conversions into POCs and on overall satisfaction rate was about 68% now, to give you some context, the global average was 37% so it’s not that different. What we’d say is actually you were a little bit more thoughtful about your decisions on proof of concepts. The success rate was almost exactly the same, four and a half versus four, so really in the same ballpark. But the key question, of course, is then, why was that conversion rate so low at the time? And also, what were the causes of that? Now, some of that conversion rate issue was because, well, it was still early, and we did a repeat of this survey back we just got back in August, and that number had gone up to launches to around seven to eight. So some of it’s a timing issue, but it is clearly part of that. Still, there’s not this huge transition going on there. And what came back, and I’ll highlight at a global level, we asked, well, what prevented you from being more successful with your AI launches? And there was, I would note, in Canada, and here you were a little bit more likely to be talking about this than other regions, but not that much more was the lack of tools for developers and others to use so that they could make full advantage and let their developers develop all these tools and capabilities. But the other thing that was showing up here and and I’m hoping that we could actually change the slides so that I can see them on the reader screen versus the notes would be to talk about, you know, there’s clearly a notice around inadequate performance of infrastructure available to do the models and extra work people want to do. But I would say the other one that came up a lot more than in other regions was actually this idea of the technology and service providers that they were working with weren’t always meeting the planned objectives, the roles. So I guess you would say, are we getting a situation where we’re seeing more promise? But when you got into reality, people were struggling with that. And then the last one, I’ll note, as we saw, pretty consistently about the same amount, is you still did expose the inner the lack of connection sometimes, between the IT organization and the business units on prioritizing the proof of concepts and what are the values and other pieces. So this is all the challenges and IDC went and we talked to CIOs and others around world and what we came up with here is this is really a pretty good list of everything that you all learned that were problems as you started experimenting more with what we called this great Gen AI scramble, which is what the last 18 months has been about. And I’m not going to go through all these, but clearly one was a sort of, did we align what we did with our strategy as a business, maybe not. We clearly heard back and again again that, you know, either Gen AI and the use of data overwhelmed any governance practices that we may have had in place and that we needed to really rethink how we think about governance of applications and data and how we exchange with customers as we do that, a number noted, yes, we have issues with maybe our people, their understanding of why we’re doing an AI initiative may not be aligning with why the senior executives thought we were doing AI executives did an AI initiatives, and that led to, basically, they wouldn’t use the new products, or they were afraid the new products were not going to be effective for them. The one I would call out here is data. We heard again and again, most consistently, from CIOs, as we talked to them, that basically they’ve discovered with all of this, that data is a dark energy in their world. It’s like we basically, we built all applications, and basically all the data in them went out to die, and we couldn’t get it back. And so overcoming dark data, as we like to call it, was one of the biggest challenges that companies came up with now as part of that. And why this is so important, why this sort of list of problems and issues to overcome is so important is that we at IDC also, though, said, all right, let’s test those comments about the impact of AI and Gen AI on the economy, not just productivity, but its actual total economic impact. And so we did actually run this process as a very standard economic modeling exercise we’ve used in the past for assessing the impact of cloud, the internet, technologies like virtualization, e commerce, and so when we roll that through, what we looked at is based on current spending now and projected spending and we’ll break out what this is the total economic impact of AI, not Gen AI, all AI by 2023, by 2030, is about $19.4 trillion that’s a lot of money, and that is on a global GDP basis, about 3.5% of global GDP now. Where did that come from? Well, the part that we’re all aware of is the money being spent by now by IT companies, cloud providers, enterprises, buying big GPU systems, paying for software with AI functions in it, actually, you know, investing in services to help implement and develop all these proof of concepts and production solutions. So that’s, that’s part that’s the driver, of the economic engine, but there’s a lot of indirect spending as well. Now part of that is the money being spent on the data centers to house all that equipment and the power that they’re consuming, and the semiconductor plants that we’re building in order to build the chips, to get those out to this and the people that we’re all training so that they’re better able to use and develop AI products. But in the long term, the impact and what for us is the big question is, how, more and more, how does AI begin to impact the economic efficiency and the business success of corporations as you embed it into your business processes, into how you engage with your customers, how you make decisions, and that’s really the next, the biggest driver of the future expansion here. And then the last big part is I should notice is like, yes, those people who are successful and involved in AI will make more money, they will have higher incomes, and they will go out and spend that money on houses and trips and education and other things. And we want to account for that as well. So you put all that together, what we would say is, right now, the promise, the hope, is that by 2030 for every dollar that people are spending on AI technology with many of the providers in the room today, it’s basically 4.6 dollars worth of economic impact, and that’s the positive view of it. We will know we didn’t calculate job loss in here. That’s not what this is about. But certainly there is an increase in some types of jobs, and we wanted to account for that. Now, this is a scenario that IDC built, and I put it straight forward, based on our assumptions about spending for AI technology and the improvements in efficiency of business that are built into calculations of sort of country level economic efficiency. And that’s what you kind of get out of this message, is that shift of now we’re going from $1 spent on AI today is accounting for about $2.4 worth of economic impact by we’re having it by 2030 that’s up to 4.6 that’s what this scenario is about. All those problems I highlighted, however, would lead to us saying, well, there’s another scenario, another option, maybe not the good option, which is that we don’t address all these challenges that we don’t really pivot more to thinking about that, and now we don’t see that same economic impact. And that’s the challenge, the risk that I’m here to talk more about and what we need to do to resolve that. So the way we talk about it is we have to move now in 2024, 25 early, 26 away from this Gen AI scramble to what we call the AI pivot, and that’s where technology providers and CIOs and C suites and technology leaders around the world go and attack the challenges and the barriers that we now have all seen that are affecting companies’ ability to take full advantage of AI. Once we do that, we’ll have also taken the steps and made some practical decisions about how to structure our AI investments, they’ll allow us to begin to plan for a massive increase in the scale of use and effectiveness of AI across all business functions in 27 and beyond, and that’s where we really see this transition to where many more and more companies are transitioning to what we would call the AI fueled business that AI is at the core of how they continue to enhance and improve their business position. So let’s dig into that. What are the key goals that we believe that companies have like yourselves have to address in the next two years to be ready for this AI pivot and to execute the pivot, as we would say. So again, I’m not going to go through all of these, but I will make a couple comments. One, it’s clear that we have to align more and more our AI strategy with our business strategy, more thoughtful decisions about where you make your AI investments that align with the business goals. And having a process for ensuring that continues is absolutely vital. The second part of that is that the other thing we like to say is that governance in this world is strategy, that governance and strategy are basically governance is the way you execute and ensure that your strategy is being effectively executed. So aligning governance and AI strategy at a business level, next critical task. Now there’s an interesting dynamic going on between the workforce and applications. So yes, every application that you’re probably going to buy in the next couple of years is going to now come with AI and AI improvements and AI extensions. But again, as we heard from a number of CIOs who did their proof of concepts, one of our biggest problem was getting the employees to use the functions on a recurring regular basis that we were trying to trial and test, and even if they weren’t perfect on the first try, that can’t just turn into them rejecting and never using it again. And we’re about to see a significant change in the application development and delivery process and how we prepare our workforce to accommodate and leverage AI more effectively through training, through understanding their needs and their requirements and how they want to see the benefits, is going to be a big part of that. And then we have a whole range of issues associated with how do we get more scale out of our AI investments, not just our Gen AI investments, with an AI platform. How do we eliminate that dark data I talked about and then at the infrastructure level, where basically everybody’s paying attention today, how do we make sure that we can get that infrastructure where we need it, when we need it, in a scalable way? Let’s dig into each of those in a little bit more detail, but before we do this, let’s step back and do a check. So I’ve highlighted what we’ve identified by talking to CIOs around the world, what they think are the seven big tasks. But where are individual CIOs and their companies on thinking about each of those tasks? So in August of this year, we went out and actually went out and asked that same 1,000 group of CI leaders now, what’s your priority for the next year across these seven different parts of the AI pivot. And not surprisingly, and when you look at it based on where they are on their current gen AI adoption, it actually is a good indicator of their thinking. We found AI platforms that idea of getting more value out of our AI investments through consistency and reuse and addressing the highly visible AI infrastructure concern that it’s too expensive or I can’t get access to it, or I don’t quite know where to put it, are at the top of the list, but we do see that for those who’ve already doing it, they are increasingly recognizing that, yes, we really need to think about the strategy part of this more. We need to get that into it in a mechanism. Now I wanted to share this included colleagues, about 107 CIOs from Canada, and we saw a different answer here. Now, not that it’s wrong. Actually, I would tell you that you’re the CIOs in Canada were actually probably being more honest, and for most CIOs more accurate on what’s going to be their biggest challenge in task in the next couple years, and that’s to understand how the infusion of AI into every application that they buy, and many of the applications that they may be building themselves, if they’re a digital business, how do they ensure that they are using that effectively, and that the people who they’ve designed it for, are using that effectively. Because the thing I will point out is, you know, while we talk about, you know, every company will be using AI infused applications. You may be having your own AI agendas, but you everybody will be using either it operation solutions or business applications that have aI infusion into them and understand what’s going on, there will be a critical task for IT leaders to address. Now let’s take that to where that leads us next. And so this is where we trivet from those tasks in the pivot to driving that scale. How do we now, once we’ve sort of done the basic ground, we’ve laid the foundation, we’ve done the groundwork, now let’s make scale happen here. And there’s really two high level goals. So first of all, when it comes to people and governance and strategy, you really do have to start thinking about at a business level, an AI fueled operational plan. How again, are we thinking about and evaluating the effectiveness and usefulness of embedding AI into business processes and solutions? And then the second part, and this is what we firmly believe, is by 2026, 27 you need to start thinking about your technology footprint in the mindset that what you’re really developing is an AI, I don’t want to say based, but an AI enabled, connected, related technology operating model. So how you aggressively you think about all your technology investments in the context of where AI fits into the conversation is a big step into it, and that deals with the applications, platforms, data and infrastructure. So let’s dig into each of these in a little bit more detail. Those two the operating plan and the operating model. So when it comes to the operating plan, as we talked about, it’s strategy, governance and people, and the easiest way to describe it is we are definitely have to move to I had a colleague who called it pilot itis is that we’ve, what we’ve had for the last year and a half has been, you know, just every pilot we could think of where enough people could say we’d have enough money to do it, we’ll go develop pilots. So we actually in a different thing we actually went out and did a survey of 3,000 IT businesses, businesses around the world, and we addressed and found people who could address their proof of concept investments by 13 different functions, marketing, sales, IT operations, security, supply chain, the whole list and where were they? How many did they do? What was their success rates? What was their investment rates? We are making that information available. There’s a great tool. You can go out and see, based on what your priorities are and where you are and what business you’re in, where that would fit. But these, across the board, were the top use cases that people identified that they’d done in 2024 to really leverage AI and so you get an idea of the different parts here. But one thing that’s become clear is, while all those use cases were what got done, that didn’t mean those were always the ones that were the most valuable or the most effective, and once again, as this, you know, colleague, you know, Linda from from Clorox, had mentioned we definitely need to take a more mature approach to proof of concepts, conversion to production, and how we align that production, because AI can’t be seen as something we just do for fun. It has to be aligned to that business value. Now, the step we would tell and what we’ve given advice to companies is we’ve analyzed those who’ve been very successful, is that they approach use cases, kind of like people talk about superfoods. A superfood is something where you get maximum benefit, not just in one part. Of the nutritional value, but across many different areas. And so there are things that provide, you know, that high value of density, high density of nutrients. They’re rich in things like antioxidants. They promote overall health. It’s time that we start applying this idea of super foods, but now to super use cases. And the way to think about this is, again, identify those use cases that build business resilience in different ways, that highlight how they have a rich business outcome and how they’re promoting the overall health of the business. And the key is that a super use case doesn’t address just one, it addresses two or all three of these. And here are some examples that we found and we saw like for those companies who really came out at the top of getting business value from their production, POC to production, transitions in three different areas. So in manufacturing, this focus on really leveraging and bringing AI in to address some specific issues that were needed in that space. The same in retail, the same in healthcare, but we are busy now identifying super use cases in many different parts of your functional business, finance, HR, so forth, but also from an industry specific basis. But that’s step one is as your senior team thinking about AI, your center of excellence and leadership. Things about where to invest AI, think about these more challenging but also much more effective and valuable solutions. So the next part of this is, as we noted, you got to align your strategy with governance, but let’s take a minute and think about what is governance? What do we mean by governance in an AI world? Well, let’s acknowledge everybody thinks initially about, oh, it’s like aligning with laws and regulations and mandates and all those are changing and other pieces in reality, yes, those influence, they shape the start of the conversation. You have to be aware of those and acknowledge them. But governance is really about again, your strategy and the systems and processes you put in place to oversight oversee the execution of those strategies. But the hard part, and what people recognize, is you can have the greatest strategy and what you think is an oversight mechanism, but if you haven’t addressed the organizational and cultural issues that basically define how people will actually do this. Will they be casual about security or data loss, or are they highly focused on it and they see it as a business value? Are all tasks? It’s only when you’ve done this and you’ve taken that first thinking about in your governance processes and thinking that you can apply that now to all the technology investments you’re making to make sure that you’re operating them in a secure way, that you’re monitoring them for continual use and effectiveness, and that you’re always assessing where you can be doing better. And it’s only by doing this and doing this process that you can really feel confident, that you can go out and claim that you are a leader in your space, in supporting and enabling responsible AI, which will be critical for regulatory and for brand purposes. And this is one clear case of it. So you know absolutely the leader for AI adoption at IKEA had made it very clear for their business AI governance responsible AI is not just a technology goal. This is a critical business value that they have to ensure that they are doing effectively. So that’s sort of the second part of it. Now it all comes back now to this AI augmented work as well. So what do we mean by this, because we’re thinking about is think through today in any organization, the workforce here the kind of the wall or the wanted poster of sort of problems with traditional work models, you know, you have the junior people do certain tasks, and that’s what they’re that’s how they build their value and their skills. You see a lot of silos of functions and operations. The C suite is kind of their task is, where do I get all the information? It used to be I had junior managers doing reports, and now I have dashboards and dashboards and dashboards and dashboards. And you see all these different things. And maybe one thing we’d say clearly, skills rarely gets to the top of the list of things that you want to think about as you do this in an AI enabled world where we’re increasingly thinking about more and more the applications, the tools that your employees are using are AI enabled, a very different set of tasks. So a lot of those repetitive tasks are about to be done by AI assistants. What does that mean for the career path of a junior employee? Integrated functions now you’re connecting different business units and business processes. Well, culturally, how open is your organization to that kind of cross organization, fertilization and cooperation? So one thing that we’ll say here again, skills development now has to come to the top of where you’re investing and understanding. It’s not just about finding new people with skills, but it’s about increasing make sure that the employees you have you are defining up front, what are the skills that are valuable, that help them do their job better and more effectively? So that’s a key step here, and that takes us now to the operating model. And the first thing I want to answer, because we always get asked is, oK, well, does this mean we’re just going to talk about, am I going to go build my large language models and foundational models? And am I going to be building this up? Are you going to do it from someplace else? And what we see is there’s clearly always a decision tree that you go through when you’re thinking about these, do I buy? Do I build? What we see most people compose. They want to. They basically take something and they want to add to it and refine it. And in that same use case survey, because we got very deep into for each of those use cases, which approach did they take for this? What we found is, is very interesting, as we noted, only about 13% of companies are really serious about building like we are going to commit to a significant investment in infrastructure on our own, our own models, our own private space. And they clearly had specific business goals, specific business values. What they thought their business was that fit with that, the vast majority would say, either I’m going to be buying, and we would actually say everybody’s going to buy. It’s just, is that the primary thing they’re going to do? Most they’re clearly, we’re at a stage, and you’re at a stage where you want to be composing that you are building more and more sophisticated, AI enabled applications that use they aren’t some of them may be Gen AI only, but most of them are using Gen AI as the new user interface to allow employees and customers and partners to take advantage of all the other advanced AI functions around prediction and, you know, and innovation and design that we’re all talking about right now. So this means that applications are going to change. So we’re going to see one of the key things you’re going to be hearing the next few years is all of your suppliers, all your are going to start talking about how they’re adding assistance to their solutions to allow you to get more out of this, to eliminate repetitive processes, to get faster responses. And pretty soon, they’re going to start talking about advisors, where, yes, we’re going to help you look at multiple contexts and multiple elements and make some advice on how you might want to anticipate a problem or or plan for new capacity or or basically eliminate a risk that was starting to appear in your organization. And then finally, and we will admit this is the last stage. We would call it the agent stage, where you actually are relying on an AI, a very complex fiscal AI system, to actually perform a significant business function. I know that one of our sponsors is SAP today, you know, and I’m not pre announcing anything, but it would be kind of like, how many companies three, four years from now are saying, you know, I’m not using s4 Hana, but I’m using Fred that, you know that that’s my accounting system, and it still is SAP, but like, you’ve changed the whole persona of what you’re delivering, how you as organizations manage this, this evolution that you’re going to start seeing in products is going to be critical to discuss. So now let’s talk about AI platforms for a second. Now us, everybody right now, is focusing on AI platforms for Gen AI. And yes, there’s a whole set of tools and things you need to do to do an AI Gen AI more effectively get better scale. But I’m going to give you a deep, dark secret. Companies already this year spent $140 billion on AI platforms. They’ve been spending it for a long time on their machine vision systems and their predictive systems and everything else. The problem before is they were all unique, bespoke solutions. What everybody wants now is scale. They want to reuse all that AI intelligence and those rules and their processes. So the big change that we’re seeing in AI platforms is unification and unification of the services and capabilities. So as you’re evaluating any company who’s coming to deal with an AI platform, always focus on, how does this help me scale? How does this enable me to do better repeatability and apply governance while doing it? And then we get to data, and we talked about dark data, and I’ll have time to go into the detail tales here, but this is one where I tell you, it’s not a technology conversation. It is a mental mind shift in your organization, where we have to move from thinking about data as a after product, a byproduct or waste product that AI in the data that’s being generated by anything you do should be approached and managed and operated like it is a product, and approaching it so that it is available and made scalable and accessible, and you’ve put in the process to do this. So this isn’t just for LexisNexis or for anybody who’s been in the data. This is for every company. Any data that you’re generating as part of your application you should always, from day one, think this will have value in the future. Let’s make sure it’s accessible and tagged and usable going forward. And then that brings us to the last part, which is AI ready infrastructure. And here we would say this is where the biggest tie between data and infrastructure is probably the next big thing everyone’s to focus on. When can I get more GPUs? When can I get other things? But no, for vast majority of companies, focusing on AI infrastructure that allows you, again, to attack this dark data problem, like don’t have infrastructure be the place where data goes to die or to molder and frame is the key next test. So absolutely, we would highlight here are the processes and steps you should be putting in place to think about data, but also the things you should be looking at from your providers of data and storage solutions and other things to really make sure infrastructure works at scale in any environment where you want to do it. So we’re coming up on the end here. I want to thank you for your time. I want to note we really, there are four steps to the pivot. So first, you really, we are moving beyond hyper experimentation with pfocs, but don’t stop doing proof of concepts. They’re still important. There’s always new developments, new capabilities. But now have in place the process to ensure that they’re aligning with their business goals and strategies. Understand right now that while everybody may be focusing on infrastructure and platforms, is the area where they kind of need to make a decision and do choices addressing these issues around dark data, and how applications are going to really start using that data, and how do we apply governance to the use of that data is really where we see the biggest opportunity and to really do things better. Last step, always take the time to consider the implications of what you’re doing on the people, on the corporate, executives on the middle executives on the staff, on your customer, because people will always make the final decision on what they think they want to do with this and what value they get. And you telling them they’re wrong is never a good strategy. So and then last I will close you. This is something we talked about, it at an eye, at a business level, but probably the part of most companies that is going to be most affected by the AI pivot is your own IT organization. That’s where all those applications, that’s where all this technology, that’s where all these changes, is going to show up first, and the lessons you learn, the things you do to ensure that your own teams are prepared to take full advantage of AI and help your customers take their customers take full advantage of AI is going to put you in the best spot to really succeed in this new world. So with that, I want to thank you all for your time. I hope you have a great rest of this conference, and I look forward to chatting with many of you today and going forward. Thank you. Thank you so much to Rick.
Transcribed by

Originally Appeared Here