Andreas Welsch: 0:21

Today we'll talk about how you can help increase your business leader's AI fluency, and who better to talk to about it than someone who helps organizations do just that. Laks Srinivasan. Hey, thank you so much for joining me. Thanks Andreas. It's great to be here. Fantastic. Hey, why don't you tell us a little bit about yourself, who you are and what you do.

Laks Srinivasan: 0:44

Yeah, I've been in the data analytics space for about 15 plus years and currently co-founder of this institute called Return on AI Institute with Tom Davenport. And, part of our mission is really helping companies and policy makers understand what this AI is and help them go through this journey faster than they would be able to do it on their own.

Andreas Welsch: 1:15

Awesome. Hey I'm, really looking forward to our conversation today and I'm super impressed that you're working with likes of Tom and other leaders as well. So if you are just joining the stream as an audience, drop a comment in the chat, what do you think AI fluency of business leaders actually entails? I'm really curious. Let's jump right in. So I, see different terms floating around when we talk about AI and readiness of leaders. Readiness is certainly one. Literacy is another. Mindset, which is the one that I favor, but I know you call it fluency. What does AI fluency entail in your definition?

Laks Srinivasan: 2:00

Whatever we call this mindset, fluency, literacy, data-driven culture is as complex as definition of AI. When we say AI and, Tom Davenport advocates for it, we use it very broadly, starting with kind of data advanced analytics, automation and everything AI. And so fluency for me is much more than answering, taking a quiz. The way we are thinking about it is how do leaders feel, think, and act about all things data, analytics, AI. If you focus it on leaders, it's about decision making, right? And decision making happens through intuition. So though we call it fluency, at the end of the day, it's part knowing some fundamentals, it's part actually applying them, and part how do you demonstrate those behaviors in a broad way as a leader. So it's a combination of those we are calling it fluency.

Andreas Welsch: 3:13

Awesome. Now I, know you know that AI has been a hype topic for the last six or seven years, and we would expect that there is already some fluency, some literacy that exists in businesses that leaders obviously have this as something that's top of their mind. But where are we in terms of AI fluency of business leaders, and especially for the audience here as an AI or CoE leader? How can you actually help and how can they help increase AI fluency of their business leaders?

Laks Srinivasan: 3:49

Yeah I, would love to hear what the audience thinks because in their own company where, it stands. But if you look at research and, there's direct and indirect evidence of where things are at direct is I remember a HBR research or article and, also an MIT article if you wanna put numbers on it. That study show 25% of CFOs and CEOs they said are digitally proficient. Digital as in data analytics, AI and I use a broader term. Boards and CEOs are about 25%. CFOs are at 12.5%. The thing that is shocking to me is 45%, only 45% of CIOs and CTOs are digitally proficient. So that's direct through some research. If you look at indirect evidence, we at the institute, we did a success factors study with Spencer Stewart, which we published. Tom and I did a webinar with MIT Sloan Management Review. What we found there is that today all points to 59-60% of the majority of the CDO, CDAO spend time educating leaders participating in data-driven culture initiatives. So it tells me the CDOs, CDAs it gets very lonely in that department. Because what they're so far ahead of the rest of the company in terms of what AI is in their fluency and literacy the rest of the organization especially the leadership is not there. So they end up going through that. I would say in summary overall the state when it comes to leaders and AI is not where it needs to be, right? I'm talking about digital non non-native, not digitally native companies. Of course, if you go into a digitally native company like Google or Facebook so that's the model, that's the target where we need to get to. But if you look at non-digital native, that's not where.

Andreas Welsch: 6:17

And why do you feel that? Is it that leaders are just so busy with other things? Or is it so complex so far out there? So little tangible this blurry thing of AI that magically solves all my problems, but I'm, struggling to apply it to my business or apply it in a context that's relevant to me. It's probably all of the above. I feel, but I'm curious what do you see what some of the

Laks Srinivasan: 6:47

reasons are? Yeah it's a great question by the way. We did a kind of one-on-one session for a CEO of a professional services company, and he comes from an industry that's used to looking at numbers. And, so he understood some of the basics very well, but you were just saying how does he spark that interest in rest of his management team? And so, why is that? If you look at it, to me, there's a number of things going on. There is a view that if they look at digitally native companies and their CEOs and how they feel they can act. They just think it's too far of a gap and therefore it's not for them. They would rather hire a CDAO. In essence, what CEOs do or leaders do is to find the talent delegate and somebody else would report the status, right? So that's part one. So that's why majority of the CDOs spend majority of their time educating leaders. And number two I think a lot of leaders are on the sidelines based on my kind of travel is because somehow they think AI fluency means they have to go learn. Take a Coursera class, learn Python, build a deep learning model. And so there's a bit of kind of misconception out there. And the way I talk to leaders who are interested in this is to say, look, you don't need to. You can still treat this as a black box. You don't need to go learn Python, build the model, but you do need to know what inputs are, what the outputs are, what is it saying. Just the CFO in a company doesn't need to be a forensic accountant. Let alone be an expert in accounting. But as a CFO O, she should know how to look at P&Ls, how to look at in balance sheet and to be able to ask them pretty deep questions about understanding risks and opportunities around a balance sheet or a P&L. So we are talking about when we say for leaders, AI fluency, we are talking at that level of at least understand enough. Because if you think you don't need to.You as a leader in a company today, you already own the risks of AI. You're using it, you just don't know about it. And clearly as a leader there are opportunities because of competitive disruption by digitally native company. So either opportunity or as risk, this is something you are in it. And therefore we think it's important especially with AI and data analytics. You understand a bit more, just enough to be able to get the organization in and lead the transformation.

Andreas Welsch: 9:47

That's awesome. Yeah I, think that resonates really, well, at least with me. And I can also see from some of the comments in the chat here. Maybe I'll pick up one. I think that today's leaders have for too long delegated digital to CIOs and CDOs, and now it's part of everyone's competency at different levels. And everyone has the responsibility to drive that change. I think that sums it up very nicely. Also what you have shared I see Jesse a asked the question in the chat. So I'll, pick that one up as well. There is a lot of misconceptions about what is required to have executive level fluency. Do you see these misconceptions? And, what do you feel they are?

Laks Srinivasan: 10:34

Yeah I, think we talked about this which is one is around leaders think AI is I'll give you an example. Part of a research, we were talking to somebody that's leading an AI project. And we wanted to understand how they interact with their business leaders. And what he said is in this company the leaders think AI, doing an AI project is like downloading an app from the app store. And so one misconception is, oh, there's cloud and there's all this stuff available. It's like clicking and downloading. Why does it take so long? So what is that? What is that misconception? So it's about unrealistic expectations. The other one is, again, I talked about it, AI is for those folks that actually build models. I don't need to worry about it. Whereas we are saying no with AI, there is some technical stuff that you do need to understand, not to go build a model, but at least you have to understand for example, most of the insights coming out of an AI algorithm, let's say it's a probabilistic output. And, by the way, business leaders deal with uncertainty every day. But, now the AI teams are talking to them about the standard deviation is, and it's a variability is this. So we are now communicating in very technical language. So, the other thing is it looks way too technical and I don't have time for it. It's other kind of misconception. And, the third one is I think somebody in the comments talks about. The millennials are reaching. So they almost think it's a generational thing as well, saying. Look, one of the hospital presidents that I interviewed and I talked to'em saying, look, Lux we are all dinosaurs and, the only way the healthcare's going to transform going to digital is as we get new leaders in. So I think that's part of the problem with this problem. If I could be bold in saying that is it's so complex. It isn't one or two or three misconception and we can put a solution. It's a very complex set of things you have to diagnose and, kinda solve and, in a sustained way for this to really make a difference.

Andreas Welsch: 13:01

Perfect. Looking at the at another comment before we move on to the next question. AI fluency has to be coupled with accepting the data is to the center. I see that in a lot of conversations that I've had with leaders and decision makers. That at the end of the day, it does come down to the data. Do you have it? Do you have access to it? Is it clean? Cleanse it somehow and, still have it in a usable form? Or how much time do you even spend getting to the data before you can get started with AI?

Laks Srinivasan: 13:34

Yeah, I could jump in on that. That's another misconception. A lot of leaders think, oh I, don't have a lot of data in my business. Or the data we have is so messy I can't really use, again, I'm using code unquote AI. And I have done a number of projects in my prior life with Fortune companies to startups. As well as through our research, what we find is companies that are generating value with AI today. They're able to put some points up on the board with whatever data that you have in whatever state that you are in. You don't need to go and say, I'm gonna put all this in a data lake and take five years and curate it and get to a data. And therefore that's another big misconception saying, what is it gonna take? How much of investment do I need to put in this before I can actually see some return? So that's another misconception that happens.

Andreas Welsch: 14:38

Yeah. I think that leads nicely maybe to the next question. And that is what's a good approach then to go firm who leads AI projects in their companies so leads the CoE and, what concrete steps can they take to bring to their leaders and to help them better understand these misconceptions or resolve those misconceptions?

Laks Srinivasan: 15:06

Yeah. I wish I had a silver bullet that you can just magically turn on. It's a long march and it's a complex. So before I get into be prescriptive about it, I think the way to think about this is with anything, right? The common framework. First of all, can I can assess the level of fluency today. Some type of diagnostic. For example, I know about this Fortune 500 company and they did a survey of all employees, including the leadership team on couple of different dimensions around AI fluency. Just start somewhere around trying to understand the current state and that would give you a lot of clues about the different misconceptions. And, so part of that is also not only some foundational knowledge about what AI is but it's also about how do I go about it? And then what are different kind of behaviors that I would've to demonstrate for me to internalize it? Meaning how do I apply it, not just learn the techniques. But how do I apply it and build that intuition that as a leader I need for me to make decisions? So first or around kinda assessing where you are and what is the state. The second would be based on that kind of figuring out some type of a program design. So one thing we find is if you make this kind of as an add-on to your project I remember some of my clients is to tell me when we do weekly project reviews and where we present insights back from their own data, it blew their minds. And some of them used to tell me saying, relax, this is the most fun 30 minutes. I have in my job learning about my own data that you as a consultant are coming and telling me, right? So, you'll have to figure out some way of sparking that interest in leaders. For some it could be playing back storytelling on things that is some that blows their mind from your own data. The second could be what I call the Ikea effect, which is involve them in some task activity. It could be one company that we researched. They talked about bringing them on in kinda labeling data in a very fun way that they need for the the machine learning models, right? Because labeling data is about business leaders saying, no, that's a good thing, or that's a bad good or bad, right, on data. So that could be another way. Another way could be I know there's Dali, there's GPT-3, I'm playing with it. They don't need to understand what it is, but they get to use it in a way that cannot gets their interest. And then the third would be you'll have to come up with some way of sustaining it in the sense that it can be just one off that happens on an ad hoc basis. It's a it's, a formal program and it could be reverse mentoring. So there's different elements that goes into a program design. And the third is how do you sustain? And, along with what we are finding is that these executives, the way they learn is not by instruction. They learn through socially meaning how do you get different executives to talk to each other? How do you get executives that have learned to showcase it as well as with a heavy dose of executive hands-on coaching? That gives them a safe space for them to ask questions without kinda putting themselves out there with the rest of the company in a way that they can actually ask questions about saying, this team came and told me. About something about R squared. What does that mean? How do I build that intuition around? Do I need to care about it? So it's a complex set of things, customizing it to your own leadership team, but then sustaining it, in my opinion, over 9-12 months before they actually develop that fluency and intuition.

Andreas Welsch: 19:11

That's a great recommendation and shows how multifaceted this challenge actually is. To your point, there's no silver bullet, right? There are different aspects that you need to work on continuously.

Laks Srinivasan: 19:29

Just think about it this way. I trained to become a firefighter a while ago, which I'm not that active in anymore. And I was just thinking about I'm not mechanically very savvy. And growing up in engineering background. But I learned, and I, was out there in a few fires and not as good as some of firefighters. I was decent, so I eventually built fire intuition around it. And so I was thinking if you look at how firefighters get trained and actually apply it, it's two methods. So one is called the drill method, which is you go to class, learn some things, but then you do every Monday night you do drills. So this is simulation and giving you a feel for it, right? And then the second is actually going out in calls and actually fighting fires or being in road motor accident. And so some of that is applying what you learn and then kinda learning what you applying and then figuring out what went well or not good and then feeding back in a sustainable way. So, part of what we may have to do here is how do you actually bring AI to teach leaders about AI? Which is there's a huge thing going on around digital twins and simulation and also personalization. So we think one of the approaches could be how do you actually bring AI and technology and in this space as well.

Andreas Welsch: 20:58

Yeah. Fantastic. What role does ethics already play in this education/ enablement around AI fluency? Do you see that the leaders are considering it, are aware of it, want to learn more about it, need to learn more about it? Where do you see ethics in that frame?

Laks Srinivasan: 21:17

It depends on first of all, ethics and everything around responsible AI. All of that is part of this fluency that leaders need to learn and there's just so many examples out there. Of things going wrong very fast and that could end up in financial, legal and reputational damage to your company. So yes there's, a lot of responsible AI and other frameworks out there. And that is part of this program. That just because you have data and you can build an algorithm, doesn't mean you go out there and do it. There are questions you ask just you would manage any other piece of initiative in your company at the leadership.

Andreas Welsch: 22:11

I think that's a good approach and a good answer to weave it into that program. Or maybe even more than just weaving it in. Making it a cornerstone of it that only becomes more important as we go forward. And, as the the scale and the reach and the impact of things like AI and AI-based decisions can influence our lives and other lives. You've talked about this being maybe a 12 month long process. Keep your leaders involved and engaged. But what happens after the 12th month? How do you know when they are AI fluent? Is there a hard KPI that you measure and you say, now they wake up in the morning and they're AI fluent, or how do you gauge them?

Laks Srinivasan: 22:57

I think there is three approaches, right? First of all, yes, there are some things that is measurable AI quotient, right? So you can have a bunch of questions that you answer. I took one just recently on LinkedIn to make myself machine learning things. That's number one. That's I think is basic. And, these leaders, they don't get into being leaders without being curious about things, being able to absorb new things and be able to repeat back. So I think that's number one. Number two, which is more very important is do they demonstrate behaviors? And, if you look at the leadership level in other things, behavior changes, that happens through a 360 degree feedback. So you have to have either the CoE or the AI teams or the leadership teams themselves trying to give feedback to each other. That's the second aspect of it. And the third, which is the most important. At the end of the day is does it actually move the needle? Which is what is happening to, as a company, is that manifesting in in, in returns on AI investments even if they don't become become a Google in a one year, do they actually go the direction of do they feel, think, and act is outta their way? Is that manifesting in real concrete evidence?

Andreas Welsch: 24:22

No not, every project is immediately successful, right? In many cases it's more like a research project than like an IT project where, you know, from start to finish that's how we'll do it. It's much more iterative. With AI, how do you convey that part to a leadership audience is especially say if, you define a roadmap you want to deliver against it, but in the roadmap, do you think that's kind of what Carlo was asking.

Laks Srinivasan: 24:50

I think AI roadmaps that's why this has to be comprehensive. What we found is there are different roles the leaders play around AI in a company, right? So there are certain leaders, could be the CEO depending on the size of the company, could be the C level teams. They're about deciding what should AI be for the company and more importantly, what should AI not be for the company? To your point, there shouldn't be science projects, right? This should be something that you declare that this is our ambition for the company. So that is one. The second role is really around what I would call a steering committee role. So a bunch of executives are in a steering role. Which is to say what is the AI roadmap strategy? And most importantly, how do I allocate capital? How do I fund this project? How do I not fund the project? How do I kill this project? So, they're in that role. The third role is a lot more around at an AI project level as a project AI leader. It could be an AI stakeholder. So they are much more around how do they communicate to AI teams data in some type of an AI language. So what we are finding is, yes, you need to educate them on all, but I would segment your leadership in these roles and then start to give the fundamentals to all of them. It's the same, but then from there, how do they apply? Depends on what role do they play when it comes to AI in the in your company. And then design a curriculum, design a program around it in a sustainable.

Andreas Welsch: 26:23

I think that's super actionable and super tangible to group them in different areas of sponsorship or how close they are to the actual zeros and ones and how close they are to the strategic aspects. I hope you in the audience find that just as valuable thinking about these categories. So hey, we're coming up on time. I was wondering if you can summarize the top three takeaways for our audience today before we wrap it up.

Laks Srinivasan: 26:55

Yeah I, think today there are a lot of misconception about what leaders under AI fluency. We think it's important that we have to get the leaders to be more engaged and become more comfortable within an organization for AI to really produce the transformational results. Otherwise, it'll just be a bolt-on under another tool in the toolbox. The third is there is a lot of content out there. But as I said, that you have to figure out what is the individual leaders, their baseline, and then create a program that's custom. And given what role they're gonna play and then put together a program in a sustainable way.

Andreas Welsch: 27:42

Fantastic. Thanks for, summarizing that. And also thanks so much for joining us today and for sharing your expertise with us. And it was really great having you on. Laks.

Laks Srinivasan: 27:54

Great to be here, Andreas. Important topic. Glad we were able to get the audience to participate as well.