AI Is Getting Cheaper. Where’s the ROI?

0:00Richard Ryffel: Mohammad mentioned processes that are going to cost people their jobs. I like to tell people, you know, you're not likely to lose your job to AI. You're likely to lose your job to somebody that knows how to use AI. So to that end, I'll put in a shameless plug here for Purdue. You know, we just passed our incoming class that just started last week. You can't graduate unless you can demonstrate certain competencies in using AI. A recent Gartner survey found only 22 % of organizations have successfully scaled AI across multiple business units or adopted an AI-first approach. Yet, 85 % plan to increase their AI spending this year.

0:40Richard Ryffel: In other words, more money is entering the system before most companies can explain how it will pay them back. Now, that certainly doesn't signal AI is a bad investment. It does mean buying AI and changing a business are two very different things. Between the investment and the payoff lies what our guests call the AIJ curve. Performance can decline as organizations learn redesigned processes and develop new capabilities. Some companies eventually climb out. Others turn the valley into a permanent address. Joining us are two leaders from Purdue University's Mitch Daniels School of Business, Rich Rifle, special advisor to the dean, and Mohamed Saifur Rahman, school chair in management.

1:19Richard Ryffel: Together, they explore why efficiency alone can't produce the transformation we're all chasing after, how AI lowers the cost of intelligence, and why the right starting point may be a use case that is big enough to matter and small enough to get done. This is the AI Proving Ground podcast from Worldwide Technology. Let's jump in.

1:45Richard Ryffel: rich just to start with you you take a look around the industry right now the capital is clearly there to support ai transformation but oftentimes we you know we read in reports or we see within client engagements the roi is not always immediately available i'm just curious from what you're seeing in the industry right now, why is it that spending is not yet translating into, you know, very vivid ROI for the companies that are investing here? Yeah, I think for two reasons. What are they doing to get ROI from AI? A first step is just trying to understand what do we really mean when we're talking about AI?

2:23Richard Ryffel: I mean, a lot of, obviously, a lot of executives are reading about this, but they just don't really know what is meant by AI. And that can be a very broad range of things. So it's understanding that, and it's educating them on how to use it. And then the next step, and maybe Mohamed will talk about this, our executive ed in particular relies on helping clients develop use cases for AI. And that's intentional so that we can leave a session with a group of executives, with them knowing how they're going to implement something, how they're going to measure its effectiveness, and actually see it through and get it done.

3:03So let's think about how this is defined for most

3:07Mohammad Saifur Rahman: firms. What is AI? It's defined mostly by popular press or what people are reading that day or what they're seeing as sort of like the main messages, right? So that's one scary part. If you are a little bit, you know, further away from that scary part. The next one is, oh, AI is all about chat GPT or Claude or Gemini or my conversation with, you know, some kind of chat bot. So for a lot of folks, right, and it's not just business leaders. Business leaders are just like most everyday folks, oftentimes, unfortunately. So they are basically thinking AI is really this chatbot. It's really more about, okay, how do I get this tool in my work processes so I can totally get the value?

4:03Mohammad Saifur Rahman: So it's a tool that lowers the cost of intelligence. That's what's happening with AI. I think if we just try to do an analogy, the internet lowered the cost of search. which is basically information gathering was easier with internet. So that changed the business models and the value that comes from it. So I don't know how many CEOs totally understand or C-suite actually understand that I'm lowering the cost of intelligence. Now, how do I really monetize that? Yeah. Right. And I think we see a lot of firms not really following this trend very thoughtfully right now.

4:40Richard Ryffel: My analogy is today we're at the point we were with the internet in like 1995, where the internet came out and every business leader said, hey, this internet thing sounds like it's the next great thing. We need to jump on this. And what you started to see in 1995 and 1996 is they basically went on the internet and posted what I call electronic brochures. Here's our website. Here's how you can find some information about our product line maybe and who works there and a contact phone number. Fast forward, you know, 30 years, and now you see the full deployment of the internet and all sorts of things that you could not possibly have envisioned.

5:21Richard Ryffel: So, you know, what we try to do with companies is to help them envision how you might apply this and to say, you know, generative AI is the tip of the iceberg. You have unique knowledge about your business, Like really challenge the people that are closest to the work to apply this less expensive intelligence, as Mohammed referenced, to making your business processes better, increasing value for clients, and ultimately increasing shareholder value. Yeah, Rich, I want to double click a little bit on the idea or the fact that AI is about lowering the cost of intelligence. What are the business implications of lowering costs of intelligence?

6:00Richard Ryffel: I don't mean that in a bad way by any means. But what does that do for business leaders and executives that need to, you know, man the ship? What do they need to know about what happens when you lower the cost of intelligence? You know, anytime you lower the cost of something that has value, you know, you can deploy it across more realms to generate value. So you could think of this as, you know, when electricity was not ubiquitous and expensive, people used it sparingly. And now you don't think twice about illuminating something as opposed to having it be operated by candlelight or a machine instead of having a human operate something.

6:38Richard Ryffel: So the implications for business, I think, are all positive. But setting those aside, I think of it this way. Back in the day, if you wanted to review your processes and see if they comported with best practices in the industry, you might hire a consulting firm to do that. And in order to bear the expense of the amazing intelligence that all the leading consulting firms had, you would have to have a business problem that was large, maybe broadly across your business, and you'd have to be able to demonstrate the ROI of paying maybe multiple millions of dollars to a consulting firm. Now you don't have to do that.

7:18Richard Ryffel: So this isn't going to put consulting firms out of business, far from it. But what it is going to allow companies to do is to find smaller problems and fix them using AI and maybe redesign some processes that they are uniquely knowledgeable of, whereas the learning curve for a consulting firm would just be so high. Think of some niche business that's doing something very unique, that it would cost me more to train the consultant on my process than the benefit I would get out of it. No longer the case. I can apply somebody internally to that problem using AI and find efficiency. So I think you'll find a lot more processes become more efficient and a lot smaller processes become efficient and really will challenge people to modernize their businesses.

8:07Richard Ryffel: And again, that's kind of the crux of what the training that Muhammad and I work on is let's come up with use cases. I'll quote one of our clients say, let's come up with a use case that is big enough to matter and small enough to get done. So these aren't necessarily moonshots, but these are things that will allow us to test the waters and get comfortable and confident in the way we apply AI to our businesses. When you're starting to identify these use cases that are big enough to matter, but small enough to accomplish and work your way through to start to gain some momentum here, how do you help organizations start to walk down that path?

8:42Mohammad Saifur Rahman: So we work with the leaders to really make sure they understand what we have to do to successfully deploy this tool where we are not really just thinking of usage like emails. So usage like just chat GPT or, you know, like Copilot or something. So rather, we take them through the foundations of what it takes for a firm to have foundational advantage in AI. Because you need to think about, okay, I'm lowering the intelligence cost. What does that mean? Well, if that means that the only KPI that I'm worried about is a bunch of headcounts, how many headcounts am I changing? That's probably the wrong KPI to worry about.

9:33Right.

9:33Mohammad Saifur Rahman: So you have to kind of worry about, OK, what are my KPIs? Well, your KPI is going to be, do I have the right kind of talent within the firm that can manage and maintain and advance my pipeline for stronger and stronger AI infrastructure? So that infrastructure involves compute for sure. Right. Not many firms are going to invest in compute. So there are compute providers. But then you have unique data, context. That's unique to your firm. That's going to be source of advantage for you. So that's your moat. Are you really advancing your moat? And deepening and widening your moat is something that you need to think as a firm very carefully.

10:16Mohammad Saifur Rahman: And the next level is, okay, how do I really think about changing my work processes? So when Rich is talking about these projects, we always talk about, okay, how do I really re-engineer this work process? Where do I really need the human intelligence still, human evaluation, right? And where do I really delegate to whatever the intelligence that I'm using? This could be the AI agents or classic machine learning models that are actually running or deep learning models that are running behind the scene calculations for me. And then really think about the KPI. I think where a lot of firms are not thinking about, and this happens with every time we see this kind of technologies taking off.

11:03Mohammad Saifur Rahman: So there's what's called a J curve. Basically, think about a J. Firms start to invest. You invest a lot. So you don't really see ROI. You're in negatives because you're making investments. And then a bunch of firms are going to just get stuck at that valley never take off. The ones that take off are the ones that actually really say, okay, you know what? I'm going to take this tool and really see if it's going to create value for me, but value that I can measure and I can justify short run, long run, right? So when you read about many firms are investing, they're not getting ROI. We're going to see a bunch of this happening just like the 2000s, a bust will happen because there is a lot of FOMO investments, right?

11:53Mohammad Saifur Rahman: Fear of missing out investments or because, you know, my board members are asking me, hey, did you invest in AI? What are you doing in AI? So the answer from the C-suite is, yes, we're investing in AI. We invested X million dollars to get this service, that service, every employee has, whatever. But you have never, if you don't really train your employees to adapt their roles and their jobs for these new tools, new capabilities. Because remember, in the past, a lot of the employees were also generating the intelligence. Their value was, so going back to something Rich was saying, their value was really knowing how to create this intelligence.

12:36Mohammad Saifur Rahman: and then go use this intelligence to monetize for the firm. And then we have to be intentional in saying, okay, some of these cases, these ROIs are going to come from a little bit longer period when I have my moat really saving me. So we take that full approach. We really kind of take learners through the whole thinking and we work with the firms to make sure we really push them to define the ROI. A lot of leaders, that's their toughest sort of challenge. They don't know how to think about the ROI when it comes to this tool.

13:14Richard Ryffel: The way I like to describe it is don't think about what you're not doing because you've applied AI. Think about what new things you're able to do because of AI. So, you know, maybe I'm increasing my growth rate without adding headcount, but I'm not just looking at it from the standpoint of how many people can I eliminate. So you have to have executives that are challenging their teams to think about what they're not doing that they could be doing that's going to drive value for their customers and their shareholders. And there are a lot of incumbents in business that resist change for a number of reasons.

13:51Richard Ryffel: Perhaps they're worried they're going to lose their jobs. So culture is an issue. The amount of work involved in bringing about change, it's a lot. And people are busy and stretched already. So do I want to add one more thing? I think of it as like a Salesforce.com implementation or a new ERP system. It's like, wow, we would love to do that, but we're swamped right now. And then the other impact, this relates to the J-curve. So the J-curve speaks to maybe the financial implications of this, but the benefits of deploying AI are latent. So if you're managing, and this kind of gets into the way people are measured and compensated, if you're managing a business and you have to deliver a certain revenue or profit target, and that's a full-time job, how quick are you going to be to say, you know what, if I invest in this project, take resources off of hitting my immediate quarterly number to do this thing, which in the long run is going to be great for me, that's a difficult, again, a kind of a cultural motivational thing.

15:02Richard Ryffel: So I think senior leaders and companies need to be thinking about their own organization. Some are more willing to step off and do something new. Others are kind of entrenched. This episode is supported by Glean. Glean is an enterprise AI search and knowledge discovery platform that instantly surfaces relevant information from across all your organization's tools and systems using proprietary AI. Unlock the power of your organizational knowledge with Glean's intelligent enterprise search platform. It sounds like no matter where you're at, you are going to most likely experience that J-curve that you talked about.

15:40Richard Ryffel: So I think the natural question after that is, well, what's the roadmap to get out of that valley? What are some of the hallmarks of either companies you've worked with that have started to emerge from the J-curve? Or what's the advice from you and your teams about what do companies and senior leadership need to do to power through that valley and start to grow?

16:00Mohammad Saifur Rahman: I think the first thing to realize is you don't go take this technology, whatever form you want it, right? Whether it's the personal assistance form or the more fancy. I think Agent TKI is the newest fashion. Everything, people just say, you know what, I'm going to put in a bunch of agents. So it's really about, do you really have a business case for the tool? You certainly do. You know, a lot of these are your pet peeves, like, oh, it's my annoyance every morning that I have to deal with all these emails. I want some agent to clean this up for me. Fine, but is it worth the compute and the time and all of that, right?

16:44Mohammad Saifur Rahman: So that's part of it. The other part of it is you will have employees, and I think this is a little bit unique compared to all other technology shifts we saw in the history, at least in my opinion. We have now the rank and file white collar workers really worried and concerned about their well-being. If that's an attitude, these people are not going to really do what they need to do to get this deployment right in the business processes to create the value. And so the incentive alignment is super important right now. But it doesn't mean that we can just stop hiring entry-level people and forget about the apprenticeship model.

17:37Mohammad Saifur Rahman: We actually kind of have these leaders, future generation leaders growing up in the firm. So I think these are the things that firms, this is where J-curve becomes a permanent place for a lot of firms that bevalley, and then eventually they fail. is because you are not really thinking of this tool as an enabler of unlocking value. It is an enabler for operational efficiency. A lot of firms are going to be pretty good at using this to save some cost because those are the low-hanging fruits. And that's your initial path toward investment and justifying them. But that is not a source of competitive advantage.

18:24Mohammad Saifur Rahman: Operational excellence is a very narrow slice of advantage for very few companies. So if you are just thinking, I'm going to get a lot of operational excellence, and that's how I'm going to get out of this J, not going to happen. Think of how do I unlock value, do new things that I have not done, change the business model. If you look at banks, they have changed their business models. with the internet. We have to change our business models with the AI. The cost of intelligence has gone down. So how do I really change my business model? And those are the ones that are going to really get on that part of the J where your value goes up like rapidly, right, exponentially, and you get the value.

19:11But it does require going beyond just operational efficiency,

19:16Mohammad Saifur Rahman: making sure your employees are assured and their incentives are aligned to use this tool for the value creation for the firm. And that's where I think education is super important. A lot of leaders don't know how to take their reports down that path because they themselves are uncertain about what to do.

19:38Richard Ryffel: I like the idea of contrasting operational efficiency all well and good, but that's not the same as operational transformation, AI transformation, which is what, to your point, is what's going to get you out of that J-curve. But that transformation needs board members, executives, senior leaders. It requires them to be familiar with not only the technology to a certain extent, but certainly the business. It's more of a balance than it used to be, where it was like the technology is IT and we're the business. It now requires expertise in both ends. So you're not going to repeal the J curve, whether that's in the aggregate across our economy or at any individual company.

20:22Richard Ryffel: But culturally, how do you get employees to embrace AI in light of this J curve? So one thing Mohammed mentioned was aligning incentives. But I also think maybe even more importantly, leaders, middle management leaders need to know that the CEO has their back. So the CEOs need to lead by example. They need to encourage experimentation kind of from the bottom up and also insist on top-down transformation. So, again, we encourage the clients that we're educating to not think of this as tech spend. And the way we've designed our executive education programs is specifically around this. And this is why business use cases are so important.

21:14Richard Ryffel: There's going to be some tech spend. You know, writ large, you see, you know, trillions of dollars being spent on AI. You're going to have to spend some money. But that's an investment in a specific business use case. So to prep for this, I did a little quick research. JP Morgan has 1 ,000 business use cases that they've identified, but 50 are priorities. Okay. So they're not just throwing stuff against the wall and seeing if it's going to stick. They're encouraging everybody to find use cases, and then they're prioritizing them. And then like any other investment, there'll be a contention system to decide how they're going to allocate the capital to those uses, and they'll deploy it to, in this case, 50 use cases.

21:59Richard Ryffel: And then they'll start to see this ROI. and then that goes viral. Other people say, hey, I want in on that. What about my use case? So that's what we're seeing and that's certainly what we encourage our clients that we work with to think about it that way. Mohamed, let's talk about workforce transformation here. We've talked a couple of times now about that here. Where do we see the workforce transforming the most? I think with AI, this has to shift.

22:29Mohammad Saifur Rahman: The shift has to be that we have to really focus and value and, I guess, nurture what I would call EQ within the firm. EQ is always valid and valued at the top level. To expect our C-suite people to be able to articulate and present very succinctly within a few minutes a big vision that everybody buys into. That's kind of the litmus test for most CEOs. But we don't expect that from mid-level managers, or we don't even expect that for my entry-level workers. We think these individual contributors, they can be in their own cubicles or in their own, I guess now with work from home, in their own basement and not talk to anybody, right?

23:21Mohammad Saifur Rahman: That mindset has to shift. You're going to be a manager of hundreds, if not thousands of workers from day one as an entry-level worker. This is what I'm telling all my students. It just happens that these workers are not human workers. These are agents. They have to know how to evaluate their output, how to prioritize whatever they're creating and say, okay, I'm going to take this. And then communicate that and see how that changes the business value that's created for the firm. So I think there has to be a deliberate attempt within the firm to say, okay, how do we really create the experiential experience of our employees faster?

24:08Mohammad Saifur Rahman: They cannot be. And this is where education and companies have to work way more tighter than ever before. Like we have to sort of get students in the classes working with real company problems. So they actually, when they go into your firm, they're ready to run. They already understand your business processes because you don't have the time like Rich's team where you take three years to get this investment bankers to sort of learn and then bring value. That three years is probably not going to be three weeks or three months. So that's only possible if they actually work through some real investment banking scenarios while they're in college.

24:51Mohammad Saifur Rahman: Similarly, your mid-level managers, they have to be thinking, OK, how do I really think about the big level, big picture contribution of the problems I'm looking at? So when he was talking about one of our clients, you know, like too big to are sorry, it's like it's a problem where it has to matter in the big picture. We always say, okay, how does this matter behind your everyday pet peeves or your everyday operational problems? Tell me why the enterprise should care about this. You'd only think about this if you're forced to think about this, even at the mid-level to a little bit like if you even think about vice presidents or associate vice presidents level.

25:36Mohammad Saifur Rahman: So I think that has to happen. And that's kind of what I think firms have to do in terms of that worker transformation.

25:44Richard Ryffel: Yeah. Rich, you've been scribbling here. What do you see on the horizon? Well, so Mohamed mentioned his students in getting some experience during our academic programs. I had a notable conversation with the head of an accounting firm who said, I need to hire students that can come in day one and basically have two years of experience on the job. And I said, okay, interesting. We do obviously through internships and practicums and whatnot give students that experience. But I said, give me a little bit more background on that. He said, well, it's a pretty simple economic thing. I used to be able to bill out an accounting grad at$250 an hour to a client to do some basic due diligence or data collection or something like that.

26:35Richard Ryffel: Now that can be done by ChatGPT and I can't put on an invoice, you know, ChatGPT$250 an hour. So I need to put them on higher level tasks. So thus one of the reasons we're tweaking the curriculum, but I would just encourage everybody, regardless of your role, you know, somebody just graduating or somebody that's been working 10 or 15 years, you need to think about it from this standpoint of how am I using AI to drive more value for my organization? Because in turn, my organization is trying to drive more value for customers and ultimately shareholders. So redesigning the process to have a human in the loop where required, because obviously if all I had was human in the loop, I wouldn't save any money there.

27:21Richard Ryffel: And then the other, I was talking, it reminded me of this old joke. You know, this person, their car wouldn't start. So they had it towed into the repair shop. And they said, hey, my car won't start. Can you help me? And the mechanic came out to the car with a hammer and popped the hood open and smacked on the engine. And lo and behold, the car started up. And the mechanic turns to the owner of the car and says, that'll be$1 ,000. And they said,$1 ,000, all you did was hit the engine with a hammer. And he said, well, hitting the engine with a hammer is free. Knowing where to hit on the engine is$1 ,000.

28:00Richard Ryffel: And so that's to Mohammed's point about this experience that you accumulate over years and years and years. You can go on ChatGPT and say, my car won't start. Where on the engine should I hit it with a hammer? What kind of hammer? How many times, et cetera, et cetera. So that's going to obviously have a massive impact on the workforce and the way the workforce is trained and the way they add value and, frankly, the way they need to manage their careers as well. Yeah. Well, Rich and Mohamed, we're running short on time here, and you guys have been gracious enough to give us a lot of time so far here.

28:34Richard Ryffel: So we'll wrap there. You know, Mohamed, Rich, thank you so much for joining us on the Ad Proving Ground podcast. Hope to bump into you guys soon, have you back on the show again. Would love to do it. Thanks for the time. Yeah, thanks for having us. been fun. Okay, thanks to Rich and Mohamed for joining. If there's one idea to take from this conversation, it's that the AI ROI problem relies significantly on how organizations lead through change. AI can make intelligence faster, cheaper, and more widely available, but it cannot decide which business problems matter most, redesign the work, reassure an anxious workforce, or tell a board which outcomes deserve investment.

29:10Richard Ryffel: This episode of the AI Proving Ground podcast was co-produced by Nas Baker and Kara Kuhn. Our audio and video engineer is John Knobloch. My name is Brian Felt. Thanks for listening, and we'll see you next time.