# Monetizing Data With AI: MIT CISR's Barb Wixom - **Show:** Me, Myself, and AI - **Published:** 18 Feb 2025 - **Source:** https://pdst.fm/e/traffic.megaphone.fm/AMMTO6917509339.mp3?updated=1743618656 - **Transcript:** https://podcasts.opennash.com/shows/me-myself-and-ai/monetizing-data-with-ai-mit-cisrs-barb-wixom-f2271f/ - **Provider:** OpenNash Podcasts (machine-generated) ## In brief Barb Wixom argues that AI should never be pursued apart from data monetization, meaning the conversion of data into real financial value. Her point is that AI is a tool within a chain that runs from data to insight to action to money. Without that tie to the income statement, projects drift into shiny-object experiments. The main tension is between speed and foundations. Generative AI shrinks project cycles from years to months and lets non-technical people use advanced tools. But many firms lack the governance, data and training behind it, which raises the risk of failure and widens the gap between winners and laggards. ## Key insights 1. **Insights sitting on a shelf create no value** Wixom says many firms build good data and dashboards, then assume users will act on them. Financial value only appears when someone changes how work is done, so the whole chain must be managed. 2. **Money matters for sustaining AI over time** Wixom separates value creation (happier staff, better customers) from value realization (financial results). She argues a use case only survives if it improves the organization's financial health. 3. **Generative AI tools and solutions need different management** A universal gen AI tool rolled out to all staff should be handled like a personal productivity tool. A gen AI solution built for part of the business should go through a traditional investment and development process. 4. **Scaling AI means three learning phases** CISR studied 52 production AI projects and found phases: making models deployable, recontextualizing them for new markets, then running processes algorithmically. Deployability required a sixth capability, AI explanation. 5. **Gen AI widens gaps between prepared and unprepared firms** Wixom worries companies use advanced gen AI while still at basic dashboard and governance maturity. Khodabandeh adds that firms with strong foundations gain faster, so winners pull away. ## Questions this episode answers ### What does data monetization actually mean? Wixom says it is not selling people's data. It means turning data into insight, acting on it, and ending up with financial value for the organization. ### Does AI success have to be measured in money? Wixom argues yes for realized value, because a project needs financial benefit to be sustained. She still welcomes other outcomes, such as happier employees or social good, as value creation. Khodabandeh adds that money is simply the easiest thing to measure. ### Why do so many company AI projects fail to deliver value? Wixom says firms often skip steps in the chain from data to insight to action to money. They produce dashboards and assume people will use them, but the insights sit unused. ### How is generative AI different from earlier machine learning projects? Wixom says it follows similar patterns but much faster, with projects taking one to three months instead of one to three years. Khodabandeh notes it also lets non-technical people work with AI directly. ### What risks does she see with companies adopting generative AI? She worries employees can easily use powerful tools while their organizations lack governance, oversight and training. She says this creates big gaps and risk. ## Facts - **Barb Wixom** [0:02:13]: "CISR just celebrated its 50th anniversary at MIT. In fact, we had the founder just give a talk, Scott Morton." - **Barb Wixom** [0:03:05]: "he started the Center for IS research in 1974. This is before we had CIOs and large organizations were struggling with managing tech and succeeding in tech." - **Barb Wixom** [0:02:13]: "he has the honor of being the very first doctoral dissertation in the area of decision support." - **Barb Wixom** [0:02:48]: "he he was trying to optimize laundry machines at the time, and his advisor said that was the stupidest thing he had ever heard." - **Barb Wixom** [0:03:05]: "CISR was established to help leaders succeed. It's actually a nonprofit, and we still do that today." - **Barb Wixom** [0:04:01]: "I have a book called Data is Everybody's Business that just came out fall of 2023, and the book was a synthesis of literally thirty years of my research." - **Barb Wixom** [0:04:01]: "So I started researching in the space of data warehousing, which, again, back then also included advanced analytics, which we now would categorize as AI in 1994." - **Barb Wixom** [0:05:13]: "Data monetization is when you start with data as an organizational resource and you end up with money and economic resource, and there's a conversion that happens in the organization." - **Barb Wixom** [0:06:55]: "the context for AI should either be changing the nature of work, changing the nature of products, or creating new informational solution revenue streams." - **Barb Wixom** [0:07:27]: "BBVA, for instance, when they first were getting really hard into AI, they did a lot of co-creation, social good work." - **Barb Wixom** [0:09:42]: "back in about 2015, we started getting questions from the leaders with whom I work very closely about data creepiness." - **Barb Wixom** [0:09:09]: "one of the five capabilities that you need is you need to have your governance and your oversight." - **Barb Wixom** [0:10:43]: "when we are rolling out a universal tool gen AI to the masses across an organization, we need to be managing that like we would manage a personal productivity kind of tool." - **Barb Wixom** [0:18:52]: "So at the time, we identified fifty two projects that had been put in production." - **Barb Wixom** [0:20:03]: "it required a six capability beyond my data monetization, five core capabilities, and it was AI explanation." - **Barb Wixom** [0:20:33]: "AI explanation we define as a capability where you can do bias remediation, decision tracing, value formulation, and boundary setting." - **Barb Wixom** [0:20:33]: "We did a large-scale case on SEMEx, where you would learn how to use AI in order to optimize a plant." - **Barb Wixom** [0:21:58]: "a machine learning project that traditionally would take one to three years with gen AI, we're seeing one to three months." - **Shervin Khodabandeh** [0:22:21]: "I've been doing AI, maybe twenty some years." - **Sam Ransbotham** [0:12:17]: "That went from, you know, sort of thirty to 40% saying that they knew what they were doing to fifty, 60% knowing what they were doing, to going back to thirty and forty percent once gen AI came out." - **Shervin Khodabandeh** [0:00:38]: "Together, MITSMR and BCG have been researching and publishing on AI since 2017, interviewing hundreds of practitioners and surveying thousands of companies" - **Barb Wixom** [0:28:37]: "So imagine I have these hundred CDOs, COOs, global, really busy people, but they patiently participate in conversations every quarter on some topic." ## Transcript **[0:00:02] Shervin Khodabandeh:** What's the most important way to measure the impact of AI on an organization's success? One AI researcher shares their opinion on today's episode. **[0:00:13] Barb Wixom:** I'm Barb Wixom from MIT CISR, and you're listening to me, myself and AI. **[0:00:19] Sam Ransbotham:** Welcome to Me, Myself and AI, a podcast on artificial intelligence and business. Each episode, we introduce you to someone innovating with AI. I'm Sam Ransbotham, professor of analytics at Boston College. I'm also the AI and business strategy guest editor at MIT Sloan Management Review. **[0:00:38] Shervin Khodabandeh:** And I'm Shervin Khodabandeh, senior partner with BCG and one of the leaders of our AI business. Together, MITSMR and BCG have been researching and publishing on AI since 2017, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and to deploy and scale AI capabilities and really transform the way organizations operate. **[0:01:05] Sam Ransbotham:** Hi listeners, thanks for joining us. The Shervin and I are talking today with Barb Wixom. She's a principal research scientist at MIT's Center for Information Systems Research, known as CISR. Barb's been leading research on technology topics at CISR since 2013 and has new research on the use of data and AI in business. As listeners know, Shervin and I have both been researching similar topics for years as well now. And we kind of feel like Barb is a kindred spirit in that respect. **[0:01:32] Sam Ransbotham:** So we're hoping to talk about some of our shared experiences and perhaps get a little bit nerdy about researching about AI in general. It's a really fast moving, complex topic that is kind of hard to understand well. And just as an aside, Barb is one of my favorite people in academia. Oh, actually, why did I even qualify that with in academia? That was dumb of me. I should have, uh, Barb is one of my favorite people. **[0:01:55] Sam Ransbotham:** So, first of all, welcome to me, myself and AI. **[0:01:59] Barb Wixom:** Oh my goodness, thank you, Sam, and thank you for such a kind introduction. **[0:02:03] Sam Ransbotham:** Let's start Barb with you giving our listeners a better idea of what your role at CISR is, what CISR does, and I gave a cursory overview. Give us some depth. **[0:02:13] Barb Wixom:** Well, it's pretty fun. CISR just celebrated its 50th anniversary at MIT. In fact, we had the founder just give a talk, Scott Morton. He came out of a Harvard PhD. Back in the 60s, early 60s, and he has the honor of being the very first doctoral dissertation in the area of decision support. In fact, he told a really funny story where he pitched as his doctoral topic the idea of taking data out of a system and using it for management information. **[0:02:48] Barb Wixom:** Basically, you know, he he was trying to optimize laundry machines at the time, and his advisor said that was the stupidest thing he had ever heard. Why would we ever waste computing power using data from systems in order to inform decision making? And Scott That didn't age said, Well, **[0:03:04] Sam Ransbotham:** very well. **[0:03:05] Barb Wixom:** no, no, it did not, it did not. So fortunately, he switched advisors, proved his first advisor wrong, and literally started the field of decision support. So when he earned his degree, he became a faculty member of MIT, and he started the Center for IS research in 1974. This is before we had CIOs and large organizations were struggling with managing tech and succeeding in tech. And so CISR was established to help leaders succeed. It's actually a nonprofit, and we still do that today. **[0:03:38] Barb Wixom:** We still help leaders succeed with technology, and my role is to help specifically with data. And, of course, now, well, it's always been AI as well, and we can talk about how I see the two of those animals. **[0:03:51] Shervin Khodabandeh:** Barb, how does your role specifically work on helping leaders succeed? I'd imagine it's through research and maybe coaching, but can you share some more specific examples? **[0:04:01] Barb Wixom:** I have a book called Data is Everybody's Business that just came out fall of 2023, and the book was a synthesis of literally thirty years of my research. So I started researching in the space of data warehousing, which, again, back then also included advanced analytics, which we now would categorize as AI in 1994. And what we found today is there's such a pervasive need across an organization, everybody from a frontline worker all the way up to a board member, to understand how to think about data and AI in productive ways. **[0:04:36] Barb Wixom:** And so that's what the book is. Well, as you can imagine, data is everybody's business. The first question is, well, how does AI figure in? And so what I've been recently doing is a lot of work to inform how AI fits into data because I, and we can debate this. I passionately believe that AI is not something you should pursue outside the context of data monetization. And I know we're gonna have to pause and define this first, because there are too many listeners out there thinking data monetization isn't that selling people data in creepy ways? **[0:05:13] Barb Wixom:** No, no, that is not how I believe we should be thinking about data monetization. Data monetization is when you start with data as an organizational resource and you end up with money and economic resource, and there's a conversion that happens in the organization. AI is a technology that helps in that conversion process. And so, for instance, AI can be used to turn data into data assets to be used to derive insight. AI can be used to derive insight, but in the end, we have to act. **[0:05:51] Barb Wixom:** That's what the mechanism is to ultimately turn into financial value. And now, even today, action is enabled by AI in terms of automation or decisioning processes kind of thing. So at the end of the day, what I care about is getting to realized value for an organization, data monetization, and again, AI is what helps us get there. **[0:06:14] Shervin Khodabandeh:** Now you talked about people should not be pursuing AI outside of data monetization, which stated differently, AI should be tied to significant value, right? Which is also, Sam, being the focus of our work. That is, if you randomly pursue these shiny objects without a strong tie to business and strategic value. If your AI efforts are not tied to your business strategy or your corporate strategy, or where you know how you are getting more efficiency or revenue or growth, then those are probably wasted. **[0:06:52] Shervin Khodabandeh:** Is that what you mean by AI should be tied It's absolutely to? **[0:06:55] Barb Wixom:** what I mean. Yeah, at the end of the day, the context for AI should either be changing the nature of work, changing the nature of products, or creating new informational solution revenue streams. Those are the three options of monetization. Now, of course, especially in large organizations, there will be use cases that don't directly contribute to the financial health of the firm. For instance, maybe you're using AI as a talent play, you know, and this happens. I see a lot in social good kinds of AI contexts. **[0:07:27] Barb Wixom:** BBVA, for instance, when they first were getting really hard into AI, they did a lot of co-creation, social good work. And what that actually did was helped attract talent into the incumbent BBVA Bank talent who wouldn't otherwise have gotten involved with a traditional kind of banking mindset type of organization. So there are absolutely occasions where selectively you would participate in AI projects for reasons like that. But the bulk of your portfolio of investments need to be somehow attached to metrics that you would see on an income statement. **[0:08:04] Shervin Khodabandeh:** What does your research inform us about that? We've seen big variability in companies across sectors, right? Their ability to get meaningful value from AI sometimes is orders of magnitude different, depending on how well it's tied to their strategy, how well they've changed the nature of work, the processes in the organization. What are some of the nuggets from your research in this aspect? **[0:08:30] Barb Wixom:** Yeah, so I'm the advocate for success. The research that I lead is both foundational as well as edgy, kind of very different animals. So the foundational work is ongoing work. And that would be what are the timeless foundations we need to be building? So what are the capabilities that are absolutely required? What practices build them? How do we design organizations and connect people to establish the data democracies that are necessary for creating again effective data monetization initiatives? **[0:09:09] Barb Wixom:** And how do we select and prioritize the right data monetization initiatives to make sure that we are making the right choices? And every once in a while, because I've been following those topics for so long, there will be a change. For instance, you know, one of the five capabilities that you need is you need to have your governance and your oversight. You could be doing great with data science, you could be doing great with your data, but if you don't have the proper oversight, you're just going to be halted. **[0:09:42] Barb Wixom:** Well, back in about 2015, we started getting questions from the leaders with whom I work very closely about data creepiness. That's how this started. And they're like, I was just told by my ex-co that I'm not supposed to be creepy. What is that? How do I manage data creepiness? All of a sudden, we realized that our current thinking about that data governance capability had to expand to not just include legal and regulatory oversight, but also values across the organization and its stakeholders, as well as ethical concerns. **[0:10:15] Barb Wixom:** Then there's also edgy, that's what I consider, like, really things that shift your mind. A good example here would be gen AI, for instance. And, you know, all of a sudden we're trekking along and we're following AI projects, and they're the traditional machine learning and natural language processing and, you know, predictive kinds of technologies. And we really think we have our handle on how these projects evolve and what it takes and what the evolution and maturity looks like for organizations. **[0:10:43] Barb Wixom:** And all of a sudden, we get hit with gen AI and it's like, what is going on here? As we were following all of their AI initiatives, all of a sudden, everybody was just realizing that the current playbook didn't work anymore when it came to gen AI. For instance, one of the latest insights that came out of the research on gen AI is the importance of distinguishing what we call gen AI tools from solutions. And basically, what that means is when we are rolling out a universal tool gen AI to the masses across an organization, we need to be managing that like we would manage a personal productivity kind of tool. **[0:11:26] Barb Wixom:** Whereas if we're creating gen AI solutions, these are more of a type of initiative, like we've always been managing, where we're investing in some kind of solution to help a part of the business. But gen AI is underlying it, and so that's a different animal with a more again traditional kind of investment process and development process in that. **[0:11:50] Sam Ransbotham:** That makes a lot of sense because if you go the other direction, you get very tool happy looking for. You know, we've all seen this rush towards gen AI, exist, everybody knows about it, let's use it for everything. And your second approach is certainly more strategy oriented. And it ties to what Shervin I found in our last research report that people were seemingly getting on top of their strategy for how they're going to incorporate AI into their business strategy. **[0:12:17] Sam Ransbotham:** That went from, you know, sort of thirty to 40% saying that they knew what they were doing to fifty, 60% knowing what they were doing, to going back to thirty and forty percent once gen AI came out. And, you know, this introduction of a new tool causes some rethinking of processes. And, you know, I want to go back to your points about monetization though, and maybe push back a little bit on that. Is everything about money? **[0:12:43] Sam Ransbotham:** I mean, some of the things that Shervin and I have been looking at are trying to find outcomes other than money. Is money the only thing? And if, you know, follow up, if it is the only thing, doesn't that lead us towards this myopic improvement of incremental edge cases with doing the same thing faster, but maybe if not the right thing? Rebuttal **[0:13:03] Barb Wixom:** Yes, it is, and here's why. First, I distinguish value creation from value realization. So just because I say it's all about money, which is value realization, that does not mean we need to be going after just financial benefits and our initiatives. In fact, I would hope that we can list myriad value creation outcomes, right, for our use cases. That would range from happier employees to delighted customers to innovative ways of thinking to more. Y- you can on and on and on. **[0:13:41] Barb Wixom:** Social good, we can start sustainability, right? However, the reason why money's important is that if you're going to set up a use case that sustains economically over time, there has to be money involved. There has to be some improvement to the financial health. Now, it doesn't necessarily mean that, for instance, again, there's a direct revenue. **[0:14:01] Shervin Khodabandeh:** Yeah, thanks for pushing back on Sam. Sometimes, **[0:14:05] Sam Ransbotham:** Shervin likes anybody that counters me. **[0:14:07] Shervin Khodabandeh:** sometimes, you know, there is a utopian view of the world that I still think that the easiest thing to measure, irrespective of how good or evil something is, actually ends up being money. And you could say for no other value than just to measure the impact, that's the easiest, I would say, variable that can be measured. **[0:14:32] Sam Ransbotham:** I guess you gave an example of a way that an organization didn't do things well. You've talked to a lot of organizations. How are they messing this up? If it's so clear to tie it to the money, if that's a great measurement tool, how are people messing this up? **[0:14:46] Barb Wixom:** Well, I didn't say it was easy. I think the problem is that in order to establish measurement, to prioritize, to make continued investments where you're building up organizational learning and ability, all of that requires really strong leadership and really strong management. You know, what I can say is that when something matters enough, you will find a way to accomplish it. Like measurement's a good example. I get so much pushback, like, we can't measure the data analytics impact to the income statement when we make this work change. **[0:15:27] Barb Wixom:** And my response is, Yes, you can. I've seen it a million times in all different ways, right? Now, obviously, your organizational context matters, like, what's the appetite for the rigor and the kind of measurement that needs to happen? There's a lot of factors that inform how well you can do it, but when there's a will, there's a way. I think most organizations don't have the commitment, frankly, to do what they really need to in order to establish the right capabilities and invest in the right practices for success. **[0:16:04] Sam Ransbotham:** I want to push on the other end of that. I'm kind of a half-empty kind of guy. What are people screwing up? You mentioned the things that people that the winners are doing. What are the other end of the spectrum? **[0:16:15] Barb Wixom:** They're screwing up the basics, which is why I have some of these basics in the book. So, for instance, the data insight action, value creation, value realization process. They screw up by just focusing on one or two components of that and not following all the way through. So one thing I don't write about are all of the failures I have sadly observed. Because I'm not like you, Sam, I'm I'm the half-full. I only want to look at the success. **[0:16:42] Barb Wixom:** Let's just show people the dream of what can be, because it it can happen. But but sure enough, yes, I have studied quite a few failures, and a lot of time it's this assumption by the organization that the rest of that process is going to happen. So, for instance, you know, I'm an organization where I work so hard in my IT group to create great data and then to produce insights, maybe in the form of dashboards with all kinds of visualizations. **[0:17:12] Barb Wixom:** And then I assume and just figure that the users out there are going to take those insights and they're going to do something with them. And in fact, most of the time, insights just sit on a shelf. I know they seem simple, some of these basics, but if someone can really sit down and trace all the way through to money, then that's half the battle for getting it right. And I'd say the ones who aren't succeeding are the ones who are ignoring that they themselves have to manage that process, or you can't rely on it just happening. **[0:17:48] Shervin Khodabandeh:** I mean, it's a good point, Barbara. I mean, it is simple, but it's hard because often what happens is finding the insight itself is challenging. But, of course, as you're saying, and we agree, is not enough. Now you have to capture it. That often requires changing massively how things are being done. What are you seeing in terms of the actual change in the trenches that's required to make this happen? **[0:18:19] Barb Wixom:** Well, first of all, you're not going to have the financial benefit without change. There always has to be change. So when we talk about action, the action means change, because you're comparing, uh, what was versus what is or what will be, right? And it's that change that creates the value. And what we discovered was it was this shift all of a sudden putting AI models, machine learning, where there was some autonomy going on into production environments. **[0:18:52] Barb Wixom:** So there was much more of this interest in incorporating AI models into operations, running core business processes, mission critical. And this was really shaking things up in terms of how do we do this effectively? So at the time, we identified fifty two projects that had been put in production. And the first thing that really made us really rethink what was happening when it came to AI is we are trying to figure out how to measure the success of these projects. **[0:19:29] Barb Wixom:** And typically, you think about more adoption, that type of measure. But what we very quickly realized is in most of the cases, it wasn't about more adoption because these initiatives were about reducing the incorporation of people in processes. And so that really was, they were like, this is really interesting. So we continued studying these projects, and then we noticed that there were these three phases of learning that each one had to go through. The first, we called it, everyone at the time also was talking about scaling. **[0:20:03] Barb Wixom:** Everyone's saying, we're going to scale, we're going to scale. We're like, Well, what do you mean by scale? Because, again, we don't mean scaling in terms of number of seats anymore. So what does scaling actually mean? All that was was figuring out how you take a machine learning based project and make it deployable. And to get it to be deployable, it required a six capability beyond my data monetization, five core capabilities, and it was AI explanation. **[0:20:33] Barb Wixom:** So AI explanation we define as a capability where you can do bias remediation, decision tracing, value formulation, and boundary setting. Then the next phase seemed to be this ability to recontextualize models. We did a large-scale case on SEMEx, where you would learn how to use AI in order to optimize a plant. Well, we have plants all around the world and in different markets. How do you then recontextualize to all these different geographies and such so that you can have the effectiveness of models? **[0:21:06] Barb Wixom:** And then finally, you had this tipping point where organizations got to a point where they were literally being driven algorithmically. Core business processes and pervasive ways were being enabled, kind of these zero-touch processing kind of situations, and we called that at scale. Organizations could industrialize in ways where they were using models across the board, using MLOps and automation anyway. So, bottom line, all of this scaling, I think, made us realize as researchers that now with AI, this really is different. **[0:21:44] Barb Wixom:** We have to be thinking about how the nature of our organizations are actually being reshaped. **[0:21:51] Shervin Khodabandeh:** And gen AI is making this even stranger than before. What are you seeing there? **[0:21:58] Barb Wixom:** So first, we don't know yet. I mean, it's interesting because we're noticing the same patterns, so the scale up, scale out at scale, we're we're noticing, but the cycle time is insanely reduced. So, for instance, a machine learning project that traditionally would take one to three years with gen AI, we're seeing one to three months. It's going to be very interesting. **[0:22:21] Shervin Khodabandeh:** Yeah, no, I think that's an interesting point and resonates quite a lot with me. So just personal experience, you know, I've been doing AI, maybe twenty some years. The earlier days, when I to put it in the terms of the cycle for value for scaling that you just talked about, right? The clunky part was recontextualization, ongoing learning, just, you know, getting it from insights to value creation, sort of the last few steps of value realization in your parlance. **[0:22:56] Shervin Khodabandeh:** That was hard because AI was very clunky. And we've all been talking about human and AI working together and different ways of working with AI and the roles human will play and the roles AI will play. That was all true, but it was very hard to do because you needed sort of super users to work with these machine learning and deep learning models or optimization models. And I see gen AI in addition to all the coolness factor and the productivity gains and all those other things that you just talked about, becoming a nice interface and removes a lot of that clunkiness from human and AI working together. **[0:23:36] Shervin Khodabandeh:** That's probably why you're seeing these cycle times be much faster because you don't need like the end user doesn't have to be technical to now interface with AI. **[0:23:48] Barb Wixom:** But this is what terrifies me, though. This is where I get really concerned. So if you think back to the five data monetization capabilities and the fact that, for one, you can't skip levels of advancement. What, in a lot of cases, has happened is gen AI, which is advanced data science. They'll start playing at that advanced level, but yet they're really, from an organizational perspective, at the level of dashboarding in acceptable data use and governance. So you have oversight risks. **[0:24:20] Barb Wixom:** So we have these huge gaps right now in organizations, and I'm concerned, I'm very concerned. And so, yes, employees can easily use these tools, but we have to make sure that their training and literacy is just as easy for them to consume because we're going to have to upskill them quick. **[0:24:43] Shervin Khodabandeh:** The beta has gotten a lot larger. And the chances of catastrophic failure because you see this sort of tool that is seemingly quite smart without having the foundations. And it can get you in trouble are a lot larger than before, right? On the other side, for the winning companies that have invested in the last two decades in a lot of these foundational capabilities. And, you know, data governance and MLOps, and, you know, human in the loop capabilities. **[0:25:18] Shervin Khodabandeh:** Those are the ones that with gen AI, and I would say there, I said, traditional AI are actually winning and getting a lot of value. So this gap is also growing much faster. Not only the, as you said, Barb, the value creation cycle has shortened, but also the gap between the winners and losers is accelerating at a larger pace. And that scares me because it sort of moves us to much more of a monopolistic situation in many, many sectors. **[0:25:50] Barb Wixom:** Yep, and again, to kind of address Sam's cynicism and give a little bit of hope, **[0:25:54] Sam Ransbotham:** oh, that's too harsh. **[0:25:55] Barb Wixom:** right? Well, I really think this doesn't have to be so hard. None of this has to be so hard if we just remember the simple concept, simple foundation. So, for instance, if you know you need these five capabilities, you need AI explanation. When it comes to AI, then when we're talking about things like human in the loop, we know where that goes. We do human in the loop, cause that helps with our AI explanation capability. We need to really invest in our data because that's gonna go into our data management capability and data platform capability to create the assets we need and such. **[0:26:28] Barb Wixom:** So when we have a vocabulary and we have some frameworks, then at least we know what needs to be. And if we focus on on those ABCs, then we really can make some magic. But you're right, gen AI is shaking things up. So **[0:26:43] Sam Ransbotham:** I'm gonna take advantage of what you said there just a second ago and put Shervin on the spot. Like, we all three try to study what's going on in organizations and study what's happening with AI. And you mentioned it was hard. Shervin, what's hard about studying this? What do you find difficult? **[0:26:58] Shervin Khodabandeh:** What's hard about studying it? **[0:26:59] Sam Ransbotham:** Yeah, or doing this research? What's hard about knowing what's happening with In the companies **[0:27:03] Shervin Khodabandeh:** addition to working with you, **[0:27:05] Sam Ransbotham:** except beyond that, everyone appreciates that, of course. **[0:27:08] Shervin Khodabandeh:** I think, I mean, look, the nature of research on AI of this kind, right? I mean, there is sort of, I would say, experiential research. When, you know, Barb, you talked about that, working with a set of companies inside out for a while, where you are at a different level of insight and understanding of what's going on. It's still maybe not statistically significant because you don't have thousands of data points, but you have tens, and it's highly informative. **[0:27:37] Shervin Khodabandeh:** I think there you really see what's going on. I think when you're talking about scaling this kind of research to thousands of companies across sectors, ultimately, a lot of this is self-reported by executives at different levels of understanding and different levels of literacy. And and so I think part of it is like, the data itself has to be taken with a grain of salt. But I find, I mean, Sam, we've been doing this for seven, eight years now. **[0:28:04] Shervin Khodabandeh:** I think in aggregate, the data across sectors is actually quite valuable and telling because you do see significant trends. And they're being corroborated by completely independent work, both like the kinds of work that CISR and others are doing. I mean, I don't know if the research itself is hard. I think capturing the value is hard. **[0:28:29] Sam Ransbotham:** I mean, Barb, you're doing this too. I mean, you mentioned speed in there. That's something that I thought might come up. What makes it hard for you to study these things? **[0:28:37] Barb Wixom:** well, I'll tell you what. It used to be hard, but what's really helped me is my data board. I can't even express how grateful I am for these leaders because so I should share the model. It's really fun, actually. So imagine I have these hundred CDOs, COOs, global, really busy people, but they patiently participate in conversations every quarter on some topic. So I pose a question. So, for instance, when gen AI was first emerging, the first question was, Okay, gen AI is accelerating you in the same direction, sending you off in new directions, something else. **[0:29:14] Barb Wixom:** What's your best use case? And again, when this first came out, they were still trying to figure out use cases. So what's your best use case? Those kinds of conversations, and then they answer, and then we talk, and we get together and really think about what insights we just discovered by that sharing. So having the frontline, very sophisticated leader, firsthand, ongoing in the research conversation, that's what now makes the magic for me. As an academic, I fell into the trap of many academics where we're always kind of trying to catch up with practice. **[0:29:48] Barb Wixom:** This is the mechanism I finally found that works to keep me at least on par with the thinking that's in the trenches. **[0:29:56] Sam Ransbotham:** Yeah, so it's a combination of of trying to understand what's going on and something that's moving really fast, as you mentioned, but also getting a lot of input and getting people. I think, I mean, you mentioned to pause and talk to you and tell you what's going on. At the same time that they're trying to figure it out happening so quickly, that seems like a bit of a challenge. Barb, it's been wonderful to talk to you. **[0:30:15] Sam Ransbotham:** Thank you for taking the time to talk with us and sharing your perspectives on this. We really appreciate it. **[0:30:19] Shervin Khodabandeh:** Yeah, thank you. **[0:30:20] Barb Wixom:** Thank you so much. What fun. **[0:30:24] Shervin Khodabandeh:** Thanks for listening today. We're back in March with season eleven. In the meantime, catch up on our older episodes and please leave us a review on Apple Podcasts or on Spotify. We'll speak with you soon. **[0:30:40] Speaker 4:** Thanks for listening to me, myself and AI. Our show is able to continue in large part due to listener support. Your streams and downloads make a big difference. If you have a moment, please consider leaving us an Apple Podcast review or a rating on Spotify and share our show with others you think might find it interesting and helpful.