Accordion

• Transcript

  Melissa Harrison (00:07):

  Welcome back to Inside Innovation, a special edition of CES Tech Talk. I'm Melissa Harrison, and the question on every leader's mind is simple. "Where is AI actually headed, and how do I use it to deliver growth?" Today, we're not just discussing productivity, we're diving into the blueprint for a competitive AI-driven future. We'll explore how companies move from AI for efficiency to AI for exponential growth, redesign their operating models, and build strategies that keep humans at the center, even as technology accelerates. To make sense of this new era, we're joined by two leaders from Ernst & Young, an organization advising the world's most influential companies on how to build trust, scale responsibly, and win. Hyong Kim, EY's Global and America's Technology, Media and Entertainment and Telecommunications Leader, and Dan Diasio, EY's Global AI Consulting Leader. Hyong and Dan, welcome to the show.

  Dan Diasio (01:02):

  Hey, Melissa, great to be here.

  Hyong Kim (01:02):

  Melissa, thanks for having us

  Melissa Harrison (01:04):

  So I want to just dive in right off the top and talk about how EY is encouraging organizations to think about AI for growth, not just productivity. I mean, how do you define that distinction, and why is it so critical right now?

  Dan Diasio (01:18):

  Yeah, Melissa, thank you so much for the opportunity to join you today. What we see in an organization, Melissa, is that it's actually three main ingredients to be able to power AI for the business. It requires the right mindset, the right skillset and the right tool set. And right now, all the capital has been pushed towards the tool set. When you invest quite a bit in the toolset, then you're naturally looking for a return on that, and that usually puts you down a path of looking for increases in productivity and costs that you can take out because the AI can do what your workers are able to do today. What we see for many organizations now at three years in is a push towards growth. And that means thinking more holistically on the mindset and the skillset required to use AI, not just to run some parts of their business, but to start to reinvent their business model and start to challenge the way that they engage with their customers.

  Melissa Harrison (02:18):

  So you mentioned there's clearly been a lot of investment around the tool set, maybe not as much in the skillset and the mindset, but I think sometimes organizations think they should just take the tool set and they should just layer it on top of everything else that they already have. And so can you walk us through how companies should think about the redesign of their operating models to have that growth mindset?

  Hyong Kim (02:40):

  Yeah. Hey, Melissa, let me chime in on that. And to your point around toolset and mindset, I think historically, when you look at the technology curves and the technology waves, technology has always been there to compliment, supplement the human for better productivity, better efficiency and things of that nature. But I think that is maybe the falsehood around AI because when you look at it as yet another tool versus it fundamentally changing how it's going to change the operating model, because this is not just about improving a function. It's not about just improving a business process. It is truly changing your operating model end to end, and therefore potentially even business model changes into other adjacencies that companies would not have other thought of. And so when you think about the journey of adoption, and adoption with an O adoption, because I think first there's awareness to Dan's point, even though we're three years into this, you'd be amazed how many people are still not aware of all the various large language models.

  (03:43):

  I said ChatGPT because that's probably the first one out to gate, Dan, but you have Claude, you have Grok, you have a number of different models that are out there and it's fit for purpose for different usages. So you have awareness, you have familiarity. We all know about ChatGPT. We all have Copilot probably in our work environments, but how are we really using these large language model tools? Are we using it just to enhance our day-to-day? Are we using it to free up time? Or are we fundamentally changing some of the things that we do on a day-to-day basis that could obviously replace the mundane administrative tasks? But then what's the next element of that? And then it is really the adaption with an A, the adaption element is like, how am I using this on a day-to-day basis? In some cases, it should absolutely do things that on my behalf.

  (04:37):

  And we could talk a little bit more about the autonomous nature of agentic AI and things of that nature. But once you've really adapted it into your business model, then real transformation's happening. And we talk about native AI companies and how you transform that. But that's a difficult journey. And Dan, as you and I have talked about within our own organization, let alone our clients, it is the humans at the center. It is the change management element, the technology is there. Then it's just a matter of how do we get people stopped from resisting those journey paths that I mentioned.

  Melissa Harrison (05:10):

  Well, and then learning how to use the technology to its full opportunity.

  Dan Diasio (05:14):

  Melissa, if I could give you a couple of examples of the difference.

  Melissa Harrison (05:18):

  I would love that.

  Dan Diasio (05:18):

  Yeah. So let's talk about software development. AI is really good at writing code. I have two options, productivity or growth. I either say all the code that I want to do, I can just use AI, do it faster and I need less software engineers. Or I can say, "Actually, I need to be a much more digital business. There's so much more software I can build now with my existing base, my software developer becomes three times more capable." Same with customer service. I have a lot of contacts and I want to take down the cost of my customer service, so now I can handle calls. That's productivity. And I start to look at using less live human agents in customer service, or you could turn that into, now I want to provide a much more proactive brand lifting, upselling capability with every inbound call, and that is the opportunity for growth.

  (06:07):

  So it's really a mindset thing, like what are we in it for? Are we in it to take cost out or are we in it to be able to really grow our brand and our business? And what we're finding with many of our clients is that when you only focus on cost, there is a ceiling to how much benefit you can get. If you focus on growth, there is no ceiling.

  Melissa Harrison (06:24):

  That's amazing.

  Hyong Kim (06:25):

  I would add to that, Melissa, around the growth agenda, and those two examples that Dan highlighted are spot on, but then let's take it to the next level in terms of what are the growth opportunities that people have not envisioned in leveraging AI. You talk about the adjacencies around a core business. We have clients who have very siloed businesses. One might have a entertainment business unit, another might have a studio, another might have a gaming. And so, sometimes when we look at these operating models, they do run them separately. And so if you have a movie that's come out, our clients are looking to see how can I create multimodal marketing content for the consumers? But then how can I generate the next game on the platform? And normally that would take many, many months, years to produce. But now with the technology at hand, to Dan's point around productivity and creativity, you can generate adjacent businesses like a game from a movie almost on the spot.

  (07:33):

  And then we talk about AI being the great equalizer and it leveling the playing field. I mean, that's an example where there's so much opportunity for new entrants as well as existing companies who obviously have a big market share in that space.

  Melissa Harrison (07:48):

  I think that's a really interesting segue into talking through the variations across industries and then just a larger global perspective, because you two both bring clients to the table that are operating across different regions of the world. So first, can you tell me just across industries, I mean, you're looking at FinTech, to manufacturing, to customer service, how are organizations rethinking their assets and their processes?

  Hyong Kim (08:15):

  Yeah. So within TMT, Melissa, and I kind of alluded to this, the TMT Industry Group is very, very broad, but they're also very connected. If you think about the technology sector, if you think about the M&E sectors and the telco sector, let me start with M&E, because this is where our clients are looking at providing the digital experiences that they provide online and transcending that to the real-world experiences on a physical level, and vice versa. You have companies who are in the cruise lines or in the theme parks, name brands that we all know who are also moving into the digital space. And this notion of the digital twin, not just from an operational perspective, but from a consumer experience, those are business models and changes that they're envisioning. And if you think about the technology platforms that's needed to execute on some of those business models, and clearly the digital infrastructure vis-a-vis the telco companies and the carriers and the operators is so critical, because no longer are consumers going to wait for latency around that experience. It's got to be real time.

  (09:30):

  And when you start thinking about the ecosystem of experiences that the consumer is looking for, so if you're online on a social media platform, we all know there's a shop or a payments or a commerce element to that. You see something, I see Dan's glasses right now is like, "Oh, I like that. How do I click and buy it?" Or whatever it might be. And then obviously, the connectivity to a fulfillment distribution platform, and it'll be on my front door tomorrow.

  Melissa Harrison (09:59):

  That's right.

  Hyong Kim (09:59):

  So, I think those are kinds of the experiences that in TMT that is top of mind for many of our clients.

  Melissa Harrison (10:06):

  I think you touched on it, but personalization is just a huge part, right? It's how quickly and what is it saying to me personally.

  Hyong Kim (10:13):

  No, that's right, Melissa. And it's interesting, I just did a webcast with one of the leaders around digital identity, and you think about personalization and how much data that you share in the hopes and the promise of personalization. And clearly, the more that you can share, the more personalized experience you're going to have. But there's also a balance, because if I'm oversharing, then I'm leaving myself vulnerable for bad actors to have data breaches on whatever companies have my data and do bad things on me and things of that nature. So, there's a balance of this whole real-time experience, as well as around personalization with data and access and privacy.

  Melissa Harrison (10:53):

  I'd like to zoom out a little bit and talk about, you work across regions, Japan, India, Europe, Latin America, clearly adoption, innovation, et cetera, is happening at different speeds across the world. In the regions that you're working in, how are you seeing that and how are they differently approaching tailoring AI strategies to local challenges?

  Dan Diasio (11:14):

  Melissa, it's fascinating to look at this on a global basis, because it is not all happening at the same speed, at the same time right now. And I'm going to do my best not to get really nerdy here and geek out on some of the tech, but-

  Melissa Harrison (11:29):

  You're in a safe space.

  Dan Diasio (11:31):

  There are some technical aspects that actually are driving some of that. For instance, generative AI and agentic AI is all based on language, like the LLM. When we think about this, it is largely based on the Internet's language. And these systems, while optimized for many popular languages, struggle with different global languages, they're not as good in Japanese as they are in English, and they're not good at some dialects in India. And that is something that definitely hurts global adoption.

  (12:05):

  There's also this element that these AI companies are generally centered in different geographies around the world. Many large countries are thinking about this through the lens of sovereign AI, and they want to actually own the ability to have full lineage of the inputs that go into the large language model that speaks the language to the citizens of that particular country. For instance, in India, there are a couple of different companies that are working on capturing the languages, the various languages that are spoken in India, and then having a national large language model that you access through chat, or you access through WhatsApp, as opposed to accessing it through an app on your smart device.

  (12:46):

  That is definitely one thing that is holding many of these countries back. And the second element that I would say is a patchwork of regulations that we still see right now. Some countries are much more active in building regulations about how and where disclosures need to be put when AI is being used, and others are playing in a battle for innovation. Those are just two factors of why there's some inconsistency in the speed, in addition to what Hyong talked about of the infrastructure and technological build out, which are clearly driving the innovation agenda.

  Hyong Kim (13:19):

  And just to add a little bit more about that, we talk about the digital divide here in the U.S. in terms of the haves and the have nots, but Dan talks about this, I would call this the digital divide around the large language models, right? Because to Dan's point, the adoption and the execution of these models in non-native English-speaking countries, to Dan's point, is slower. And as I've traveled and spoken with many of our clients throughout the world, you can see the different pace of adoption that's happening because of that. And when you think about a global company that needs to have operations in the U.S. and in some of these other countries that's dependent on these large language models, it's a tricky operating model to incorporate, because hallucinations, false positives, bias and all that will happen. And so, just because you execute that model in the U.S. and you get one result, you may not get the same result in Japan or in Africa or India or elsewhere because of the translation to English to that country's native language is very difficult.

  (14:26):

  And so, there's studies out there that continues to look at this, but that's a huge consideration for many of our global clients as well as local clients that are in these other countries that may not have the English-based large language models as our native tool.

  Melissa Harrison (14:44):

  We've talked a little bit about some of the issues that global companies are facing from an adoption standpoint. I love to give a little hope nugget to our listeners. Do you have some examples of how AI is working really well for these companies and how it's driving growth?

  Hyong Kim (15:00):

  Yeah. I mean, we touched a little bit about some of the productivity, and I think that's where in isolation for their set, and when I say set, talk about their data set and their localized economy, we're seeing progress there. But then it's just a matter of, are there large language models that in those countries that can compete with the Claudes, the Groks, the ChatGPTs, et cetera, to advance some of the thinking and advance some of the autonomous nature of where this is going. Because soon, we're talking about ChatGPT and we're talking about AI, the next evolution's going to be physical AI and devices and things of that nature. And so, I think our clients in these countries are doing well on the productivity piece, but the revenue elements of pushing the boundaries, I think there's some work to be done there.

  Melissa Harrison (15:54):

  Dan, you mentioned earlier about how EY emphasizes putting humans at the center of AI. How do leaders foster a culture that embraces AI while reshaping roles and mindsets for the future?

  Dan Diasio (16:06):

  You're kind of hinting towards the elephant in the room, Melissa. And this is that while people are very excited, while workers are very excited about agentic AI and it taking what we at EY call the toil, like the stuff they really don't want to do, very excited to take that out of the job, there is a considerable number of workers around the world in all industries that are really worried about AI replacing their job. And I'd say this, I think we can all acknowledge that this is a big technological revolution. In the beginning of many of these technological revolutions there's always this belief and fear in the beginning, and we end in a period of reinvention.

  (16:44):

  What we see companies that are successful at putting humans in the center are really articulating a couple of different things. First, they are adopting more of a growth agenda. That growth agenda of investing in the skillsets and the mindsets of their workers is helping to drive a more clear agenda and a strategy of what they hope to achieve and is bringing them along for the journey. So, they start to see how their roles will evolve and how they're not slowly being chipped away and being asked on how much time they saved by using this tool from their normal day and work.

  (17:20):

  The second thing that is really fascinating is they're starting to find with just using AI puts you down a path of very statistical sameness. It's like all average. So, companies that want to differentiate themselves are recognizing that AI lifts the floor, but it's humans that actually break through the ceiling. And they're investing in building out the skills that people have to be able to work together with the AI system so they can create true differentiation.

  (17:52):

  I think where we're starting to see is that instead of the narrative being that there is a worker that does work today and now they will be replaced by AI, instead the narrative is, that worker will now have a bunch of AI systems working for them, and they will move their job from somebody that is executing processes to somebody that is starting to create new capabilities inside an organization. And all this really articulates into what the company's strategy is. So if a company's strategy is about growth, they're talking about how we all go on this journey and start to go tackle new products, new services, new business models. If the focus is really just on productivity, then it becomes harder to incentivize and motivate many of the workers to be able to go through that journey if there's not a clear definition of where it ends.

  Melissa Harrison (18:43):

  So is bringing the employees along on that journey the biggest barrier to organizations that are trying to shift AI from productivity to growth

  Dan Diasio (18:53):

  It's setting a strategy with intent and writing that intent down and making it clear across the organization. And when you do that, that results in different training that you're giving to your employees, and different roles and jobs that you're creating for the employees to move towards.

  Hyong Kim (19:11):

  I think the human element is the biggest resistance to change in this. And to the earlier point of where the technology waves and the technology curves have always helped the human, if you will, by it being a tool for better efficiency and better productivity, GenAI and agents are different, right? And to Dan's point, in some cases they will replace what humans are doing. But I think we need to look at the broader picture, because right now there are things that humans can only do and there are things that the agents will do much better, much faster. And we need to figure out what that balance is.

  (19:52):

  And I'll give you an example. And again, not to necessarily geek out too much, but I'll give us a simple example around third-party risk management. Third-party risk management, Melissa, is where our clients are looking to do due diligence on their third-party vendors and partnerships. So, whether it's financial risk, whether it's reputational risk, whether it's regulatory risk, whether it's the executives and the employees' risk, right? There's a number of things that we have to look at.

  (20:22):

  Well, historically, we would have one very limited number of data sets that we could look at both internally and externally. And it would be a subset, because humans can't process gazillion rows of data, and so it would be subsets of data. But now with AI we're going to have a much more robust data set, a much more enriched data set to give you a better answer. So, how does that help the human though? Because I can do better, faster. Well, because there's always going to be exceptions and all the things that we do, whether it's audit, tax, third-party management, due diligence, there's always going to be exceptions. So, the more third-party risk management, due diligence that we do, the more exceptions are going to fall out.

  (21:07):

  Even though you have a better result, there's going to be more exceptions, and you still need humans to look at those exceptions, because they need to determine that the model kicked it out, gave some reason that this is a high, medium, low risk, and then someone needs to determine, was that right? And so, I look at it as, yes, there's going to be things that to use dance words, the humans are going to be the toils that agents are going to do, but the volume of work and the kind of work is going to be different and we're going to need humans at the center for all of that.

  Melissa Harrison (21:40):

  I think what's so interesting around the conversation we're having is that there is this narrative and misunderstanding that it's human versus AI rather than human and AI. And your research clearly shows that organizations are actually often reinvesting those productivity gains in growth initiatives, rather than just reducing headcount. How are companies approaching this balance in real time and real practice?

  Dan Diasio (22:07):

  Yeah. What I would say is we see many of our clients as they are looking to slowly find ways to be able to improve the way a process works. They're also going through an effort to reinvent the entire way that that function operates from the ground up. I think chipping away, if you strand out, let's use an example, let's pick a space that is billing, something bland, but everybody can imagine billing. If billing takes 56 steps, you can imagine in billing that there are a couple of steps that are really conducive to AI. So, in the past, the target has been, go find those areas and build use cases around it. What we see many of our clients now, they've been doing that and they've been getting some good benefit in that, but now they're starting to shift to say, that's been helpful, but again, it's been overly focused on productivity. Now I want to turn my billing function into something that is AI native.

  (23:10):

  Imagine we throw the 50-something steps that we have away and we start to rebuild them from the entire bottoms up. That requires there be new jobs for people in these roles. That means that there are new controls that would need to exist in this space. There is new technology that is required, because it's not just an agent, but now it's something that is operating in a fairly autonomous way. We see companies really making this now shift towards the transformation of their business. And the productivity gains are helping them start to figure out how they invest and transform their workforce to those new jobs of the future.

  Hyong Kim (23:48):

  On that point, and you touched on it earlier, as these changes in the operating model and the changes in the workloads, if I could use that, for the employees change on their day-to-day responsibilities, I think there is a era of where humans obviously will be managing agents. And how agents and employees coexist is going to be, I think, a new paradigm shift that everyone is going to have to get on board with, because regardless of what level you are in an organization, you could be a person right out of college or somebody with 30 years of experience are going to have to know how to manage agents and employees together to get the outcomes that we've been talking about.

  Dan Diasio (24:34):

  Yeah. Hyong, I totally agree. I think we are coming up on one of the biggest transformations and revolutions in management that we've ever seen, which is like the corporate ladder used to be, you take a job, you work, you gather good experience and with that experience, when you've gotten really good at doing the job, you get the managed people that do the job. Now coming out of school, that's like completely flipped upside down. You are a manager on day one and you have to be a manager of maybe a human workforce, but also an AI workforce, an AI workforce that is extremely powerful, but sometimes clumsy, and those are skills that need to be really invented and developed for people.

  Hyong Kim (25:17):

  I think this is particularly important for those who are in college and universities thinking, "Oh my God, how am I going to get a job?" And so I think the skillsets that you're learning in school, whether it's a hardcore STEM degree, I think it's even more important to be able to augment that with the humanities. Because we all know the better the question, the better the answer, particularly with these large language models. And so, your ability to ask the right questions in the way that you need to so that you can respond is important. But then obviously, the results that come, whether it's something that has to do with the liberal arts, the humanities or even a technical, you still have to have that technical knowledge to vet that is the right answer.

  (26:00):

  I think not just the management elements that we discuss, it's the core curriculum that I think students need to think about. And that carries over to people who've already graduated who are in the workforce. So, no longer are you the core expert. You need the other soft skills and the management skills to advance in your career.

  Melissa Harrison (26:21):

  I actually sit on the university board of trustees for my alma mater, and we just had this conversation about how important the critical thinking skills are to be able to utilize AI to its fullest. Because it is exactly what you just said, garbage in, garbage out. So you got to be able to know how to approach it appropriately as well. Well, I'm going to now ask you maybe the hardest question of the entire podcast, because I'm going to ask you to narrow it down to just one thing. For companies that are starting on their AI growth journey, what is the one thing that you'd recommend they focus on first?

  Hyong Kim (26:57):

  Wow, I think it has to be getting people on the journey. The change management elements and the humans at the center elements of our discussion that we had, Melissa, is so critical. If you don't bring your employees on the journey and let them know how they are a part of it, it is going to fail. And so we know the technology is there, but the resistance to the adoption transformation is absolutely the human pieces, so they have to be a part of that journey.

  Melissa Harrison (27:28):

  Okay, Dan, it's your turn.

  Dan Diasio (27:31):

  I'm going to cheat. I'm going to agree with Hyong, his answer, because there is no company that reinvents itself without its people. Full stop.

  Melissa Harrison (27:40):

  I know we've just had a really interesting conversation. Anything that we did not cover that we would be remiss in not sharing with our listeners?

  Hyong Kim (27:50):

  Yeah. Dan, I think one of the things that would be as we start to think about the advancements of agents and autonomous decisioning, we did talk about how do agents then market, sell and service to agents? How do agents work with other agents? And so, this complexity of humans to agents, agents to humans, agents to agents is something that we need to think about. And in the context of the earlier point around regulation, right? Obviously, regulation is there for a reason to protect the consumer, to protect economies, to protect capital markets, but it can also stifle some of the innovation that's happening there. And when you think about where all of this is going, it's pretty exciting. And so, my message, I think to the audience is change is inevitable, adopt it and good things will happen.

  Dan Diasio (28:46):

  One more add. In an example of shifting the mindset from productivity to growth, let's describe what Hyong was describing. Let's pick the area of marketing. Today, the objective is to really optimize marketing, and I was having this fascinating conversation with the head of marketing for a consumer products company. They were describing how we could use AI to get better copy, get better product descriptions and even start to use some specially tuned AI to be able to start to develop and produce videos and images of their product from a marketing perspective. And it was all very much focused on today and how we can use those capabilities today to market in the same way. But we started to really challenge ourselves. If we're building a capability now for the future, what do we expect our marketing function to need to be three years out, or five years out?

  (29:44):

  And in this particular instance, this client and us were pretty bullish that AI and agentic AI is going to start to grow pretty significantly. And we got to the point where we realized if you want to really build a great marketing capability, you're no longer you're going to be marketing to people. In a lot of cases, you're going to be marketing your products to people's agents. And instead of having a customer relationship platform, we need to start to build out an agent relationship platform. So, the agents could start to see these products get recommended over what their competitive products are.

  (30:17):

  And that starts with the question. So, I think often starting to look into the future and really challenge what is the business that we're going to be in a couple of years, that will start to inform the growth agenda and where we start to move from a strategy perspective.

  (30:32):

  So, we don't just get caught automating what we do today. The second point that I want to mention, and this is completely out of sequence, part of our experience in learning what works and what doesn't work for our clients is by applying this internally in EY's organization. If you make a list of all the industries or organizations that are going to have their business models get impacted by AI, professional services is near the top of that. So, we've been really proud in our organization's push towards setting a very specific goal centered on trust and centered on humans, but to really not just enable our business with AI and power that business with AI. And we're also really proud to have just recently joined Microsoft and Harvard Business School's collaboration on the Frontier Firm Initiative. Because what we are going to be doing inside our organization is really researching deep concepts of what skills do we need to build out to have people operate as agent bosses?

  (31:42):

  And what are the skills that really define the human and AI collaboration? And the great thing in working with Harvard and other universities is then we can start to pass that down the education system and start to make sure that we are getting a workforce that is ready to be managers on the first day of their job.

  Melissa Harrison (32:00):

  All right. Final question. What do you expect to see at CES, and do you have any thoughts on what might dominate the conversation?

  Hyong Kim (32:08):

  Yeah. Last year CES was mind-blowing. I think this CES is going to be more around physical AI, Melissa. We've talked and heard about obviously all the great things around large language models in the software space, but physical AI and AI on the edge, I think we'll see a lot more agentic AI enabled devices, I'm hoping at CES.

  Melissa Harrison (32:33):

  Dan, what do you think we're going to see?

  Dan Diasio (32:35):

  Totally agree. See you in the fall.

  (32:39):

  Next one answering first, Hyong.

  Melissa Harrison (32:41):

  I can't wait to see both of you in Las Vegas, and we'll see if your predictions are correct. Hyong and Dan, thank you both for joining us and thanks to everyone listening. That's our show for now, but there's always more tech to talk about. Be sure to follow, subscribe, like, comment and whatever else you need to do to keep those algorithms happy. You can get even more CES and prepare for Vegas at CES.tech. That's C-E-S dot T-E-C-H. Our show is produced by Nicole Vidovich with help from Paige Morris and Doug Weinbaum, recorded by Andrew Linn, and edited by Third Spoon. I'm Melissa Harrison on this special inside innovation episode of CES Tech Talk.