CDOIQ Symposium 2024 | Mario Faria | Carnegie Mellon University

theCUBE.net

Clips
News
More from CDOIQ Symposium 2024

Mario Faria

Adjunct Professor

Carnegie Mellon University

Thoughts on CDO role evolution amid an evolving landscape of AI

Generative artificial intelligence has upended every responsibility in the technology world and is making a big impact on chief data officer role evolution. Where the first chief data officers were just involved in areas such as data quality, they’re now embracing AI.Some might yearn for a world in which one executive is tasked with data analytics, AI and company transformation. But some organizations might have one chief data officer, another chief AI officer and a CIO, all of whom may not always play well together, according to Mario Faria (pictured), Chief Data and AI Officer Program professor at Carnegie Mellon University

play_circle_outline Importance of integrating data and AI roles for organizational success

play_circle_outline Balancing technical acumen with people skills and culture building

play_circle_outline Considerations for appointing a Chief AI Officer based on organizational readiness

play_circle_outline Challenges around governance, compliance, and legal aspects in AI implementation

Info
Transcript

Mario Faria | Carnegie Mellon University

Mario Faria

Adjunct Professor Carnegie Mellon University

Mario Faria, an adjunct professor at CMU and longtime CUBE alum, shared his journey in the data field from the early days of the CDO trend to the present. He emphasized the importance of culture in driving successful data initiatives and highlighted the need for organizations to start small with AI projects to demonstrate quick returns. Faria also discussed the evolving role of the chief AI officer and the integration of data and AI functions within organizations. The conversation touched on the challenges of AI implementation, including legal and compliance ... Read more

explore Keep Exploring

What is the importance of understanding and defining the roles of chief data officers, chief AI officers, and CIOs in organizations and how they can work together effectively? add

What qualities have the best chief data officers and CIOs demonstrated in order to be successful? add

What factors should a company consider before deciding whether or not to hire a chief AI officer? add

What factors do people often cite as reasons for not being able to get into production, and why is it important for the chief data officer to be involved in the AI strategy? add

bolt Powered by CUBE AI

Mario Faria | Carnegie Mellon University

search

>> Hello, everyone. Welcome back to CDOIQ in Cambridge, Massachusetts. My name is Dave Vellante, and I'm here with my co-host, Sanjeev Mohan. Mario Faria is here. He's adjunct professor at CMU, longtime CUBE alum, back to 2013, Mario. I think the very first CDO we ever interviewed on theCUBE. So congratulations to us for getting such a great guest, and congratulations to you in an amazing career. It's so wonderful to see you again.

Mario Faria

>> Thank you so much, Dave, Sanjeev. It's been a pleasure. I know although we have not talked since 2013, I'm a fan of your program. I always follow what you do in the conference. One, thank you for helping the data community. Thanks so much.

>> Oh, thank you so much for saying that. Now you are a veteran of this conference. You remember the early days. We started in 2013. You've been here for many, many years. Really started this, helped start the CDO trend, educate people on what it all meant. What's your journey been like? Share with the audience.

Mario Faria

>> Okay. So the first time that I was here was 2011, and we had 29 people on that conference.

>> 29 people?

Mario Faria

>> 29 people. We could fit a room like the one we are in right now. Most sponsors nothing, and we have evolved so much because now we have 2,000 people attending this conference. So when we talk in 2013, I was with The Gates Foundation. After that, I joined Gartner, which I met Sanjeev-

>> We were colleagues. Yeah....

Mario Faria

>> We were colleagues at Gartner. I spent one year at AWS, and now I'm a professor at Carnegie Mellon. I'm advisor of two companies. I help some people on their careers to become CDOs, and also now I would do some consulting job.

>> What are you teaching at Carnegie Mellon? Share us the curriculum.

Mario Faria

>> Four years ago, Carnegie Mellon started a chief data officer program. I was invited to be part of that program, and I'm teaching data products, data monetization, data strategy, and it's been an incredible journey. And we are just releasing a new program that in the next month, instead of just being a chief data officer program, it's going to be chief data and AI officer program.

>> What's interesting, so many questions come out of that, Sanjeev. But so when you talk about data products and data monetization, the conversation years ago when you started, it was data quality, a back office function. Now it's about making money, driving productivity, I'm sure. And then the other piece of that that was interesting is we were talking earlier about CDO, chief AI officer. Do those two roles come together? Of course it depends, but oftentimes they should. And so the fact that you're actually launching a curriculum around that is quite fascinating. Explain the rationale there.

Mario Faria

>> Yes. So we have to understand, we live through constant change. So this world is evolving. You mentioned the first chief data officers were just involved in the backups, risk, data quality, and then they start doing analytics and digital transformation. And nowadays, they're embracing AI. In a perfect world, yeah, we have just one executive responsible for data analytics and AI and of course, helping the transformation of companies, but the world's more complex than that. So in some organizations, we have one chief data officer, another chief AI officer, and then a CIO, and how those individuals play together. Sometimes well, sometimes they do not. So what we try to teach in our program is about that, how the role should be, the main responsibilities, how the leaders should be putting a plan together that's driving companies on their mission to reduce costs or gain more money or improve productivity, and most important, have a less risky approach to AI.

>> Okay. I'm interested in your thoughts. We've been talking a lot about, Sanjeev and I were semi-debating. Should a AI leader be a technical leader? Should that person be somebody who's more curious? How do you see it? Talk about the importance of building culture, but at the same time, AI is so new. You can fool a lot of people with magic. And you can be-

Mario Faria

>> And buzzwords....

>> and buzzwords and be allured by that, so I appreciate your point of view. How do you see that balance between technical acumen and people skills, culture building, business skills? Is it a two tool star? How do you recommend organizations approach that decision?

Mario Faria

>> Yeah. Being technical helps, but it's not mandatory. The best chief data officers and CIOs that have met in my life, they did not have a technical background. They were able to understand. They were able to put together an organization with skills that would help the company to achieve their results. So you talk about culture. Culture more and more is a tremendous lever. At the same time, obstacle for any data leader to achieve success. Data is abstract. We all agree about that. And when we talk about AI, it's even more abstract than that. So we're talking about all those tools available there, all those algorithms and solutions out there. But the point is what does it mean for a company to embrace that? What are the steps that they need to do? Is the company prepared to do everything? So culture is a tremendous topic that you have to bear in mind. This is one of the reasons that I'm here at this conference to deliver a session on the good, the bad, and the ugly of building a data culture. We all know that's important. We all know what we need to do, but how to do it. That's what we're going to be exploring here in my session.

>> See, this is very interesting because we were discussing earlier about... I was more leaning towards technical capability because I come from this hands-on practitioner side. So data culture and all these soft topics to me are not-

>> They're soft topics.

>> They're soft topics, yes. Yes. The culture is built into the DNA of the leader. And it's not something that you just say, "Tomorrow, I'm going to have this culture." I don't know how people develop a new data culture, and what does it even mean? And how do you sustain it?

Mario Faria

>> Yeah.

>> So...

Mario Faria

>> The point is the data culture is quite intrinsic to the company's culture.

>> Yes.

Mario Faria

>> So you cannot build a data culture unless you understand the current moment of the culture of your company. Are the leaders prepared for a change? Do you need to do extensive program? How change management will take a role on that transformation? So although it's soft, it can really bite you if you do not take care of it because you're dealing with human beings.

Mario Faria

>> Yes. And that's the most complex thing in the world is dealing with people. But what we are saying is that if the CEO or the chairman or the highest level has a right culture mindset, they can push it down. But if a data leader comes in and says, "We need a cultural change." That's very hard to push it up and then push it back down to the rest of the organization.

Mario Faria

>> That's correct. Even if a CEO sometimes tries to push down, he or she might not be able to do that because a lot of things are so intrinsic there. A lot of old behaviors, and I like to say culture is what you do when the doors are closed and the light is off.

>> Ah.

Mario Faria

>> Okay, okay.

>> This is an interesting discussion because culture can just be an organic thing that emerges from an organization, and it is almost always, but it also can be something that comes from... You were at AWS. Those leadership principles, they take them very seriously. And you either embrace them, or you shouldn't be at AWS. My old company at IDG, Pat McGovern had 10 things, and he would sometimes quiz us on them. We used to think, "Oh, come on. Soft, fuzzy stuff." But then it became a decentralized organization. It was a mantra, and it actually became... So your point as a question is a very good one. How do you maintain that? So my sense is it either happens organically, like it does in our company. You guys know, we try to lead by example. We work really hard. We expect everybody to work hard and grind. We text them on weekends, and it's not for everybody. Others have very disciplined, well-thought-out list of items that they make part of their ideology. And sometimes it works. Sometimes it's like, come on. I'm not going to buy this. It's very hard, like a football coach has to get people to buy in or a soccer coach to buy to the message. Very tricky thing, culture.

>> I have seen in my past leaders of my group or company come into a meeting with Harvard Business Journal with yellow sticky notes. And you can tell they just read an article about culture, and now all of a sudden at the all hands, they're like, "This is going to be our new culture." It doesn't work.

>> Yeah, no. That doesn't work. .

Mario Faria

>> I like that about the principles that will guide the changes, the principles will help the company to scale. So if you bring new people, if some people retire. So the principles will allow you to move with the culture to make sure that everything's ingrained with the people who work there.

>> Interesting.

Mario Faria

>> Yeah.

>> I want to come back to the chief AI officer. When should a company have a Chief AI officer? When should it not? Give us a framework to think about that because I know I've talked to some CEOs. I haven't talked directly to Chuck Robbins about this, but I've talked to people who work for Chuck Robbins at Cisco. They don't have a chief AI officer, and it's not right for them. They just burst over. But then it's a company like Dell who, Jeff Boudreau, they put him in charge of AI. And so there's two different philosophies. Both can work. How should we think about whether or not to have a chief AI, or should we always have a chief AI officer?

Mario Faria

>> Or maybe that responsibility should be laid upon the CDO, or it may even to the CIO. After we had our conversation a few years after, I've met one individual in a conference that came to me after my presentation, asked me, "Mario, does my company need to hire a chief data officer?" And I gave the consulting answer, "Well, it depends." So we talked a little bit. And after talked to that individual for 15 minutes, I told him, "No. You don't need a chief data officer now because you're doing that work. Maybe you need to empower your data organization, bring people, but they should be reporting to you because you're doing the right things." So coming back to your question, Dave, should a company hire a chief AI officer? It depends. Are they already doing the right things with data, good quality processes, good governance? Are they exploring analytics at scale? If so, maybe they're ready to take the jump and appoint someone being responsible for AI. Maybe they're not doing the basic things, so they should be doing a lot of groundwork, the foundation that will allow them to explore AI further.

>> And I personally think it makes... Again, this always depends, but to me, the chief data plus AI officer makes a lot of sense. And I would put governance as part of that and not part of data management. And so that to me is a very sensible, and you're right. It depends on the scale, no doubt. I mean, there may be smaller organizations who need a dedicated technical leader in AI to really go deep because they can drive unique competitive advantage from building technology versus applying technology. But I think that putting those together, so I think Databricks with their conference name has that right. They call it the Data plus AI Summit-

Mario Faria

>> Summit.

>> They've co-opted that.

Mario Faria

>> Right.

>> And you know what I find really... I'm actually smiling to myself because I heard you say the answer is it depends, and then you use it depends. And something that you may not know today is exactly the third anniversary since I left Gartner.

>> Congratulations-

>> Thank you....

>> on a great decision.

>> Yes. And I created my podcast called It Depends.

>> That's like John Ford on the one hand.

Mario Faria

>> And one thing about AI, a message that I always like to tell the other chief data officers, the other data analytics leaders out there. Please do not over complicate AI. AI is a technique with a lot of tools. It's great. It's revolutionizing, but AI is there to solve some problems that are boring problems, I would say. So you're going to be solving problems with your call center, with your logistics, with your warehouse, everything about customer retention. So try to look at the boring problems that your company have, see how you can apply some AI techniques to help them out. Don't try to boil the ocean initially because you're not going to be able to do that, and you might bring yourself to trouble. So maybe your company's not prepared for something fancy, but definitely for something that you can apply some proof of concept, some problem that's out there that nobody was ever able to fix. That maybe with some AI algorithm, you're going to be able to do that.

>> So but to that point, you look at IT spending right now, it's probably growing in 3.5%. Not crazy. Back in the pandemic it was 6%, 7%. It was much higher. GDP is maybe, let's call it 1.5% to 2%. When we look with our partner ETR, we do these surveys. They show 40+% of the customers are stealing budget from other places to fund AI. We're seeing them push out now slightly the expectations for ROI, so I like to say they're hitting singles, a baseball analogy. They're not hitting grand slams and home runs. They're getting little wins on maybe productivity, a lot of experimentation going on, and feels like there's a bit of a backlash now. A lot of less hype. You look at Nvidia stock, you think, "Oh wow. AI's making everybody money." It's not. It's not throwing off enough cash to be self-funding now. So how do you see that with the organizations that you work with? At what point does AI actually have to demonstrate monetization and be self-funding? And what happens if it's not?

Mario Faria

>> Yeah. I think from the initial point, from moment zero, you have to start putting something together that you can see a return on that. The majority of the companies out there, they're not Nvidia, they're not Microsoft, they're not Amazon. They're real companies with real problems. So when I was at Gartner, worked with a healthcare provider and to implement a solution that use AI techniques to detect skin cancer in some exams that they were doing. It was not something that was replaced the doctors. It was not something that it would work there without the doctor intervention. The AI algorithm will allow the doctor to really look for patterns that he or she might not be seeing at one point. So my recommendation, start for something that you can show the benefit right away because it will help you to increase your budget, to give you more power to do some fancy stuff. In the beginning, don't try to be fancy. Do something that's very small that your organization, your team, you all can learn from that experiment.

>> And of course, when you ask people why are they not able to get into production? They cite legal, compliance. It's all those governance things. This is why I come back to maybe the chief data officer has to be exceedingly involved in that AI strategy.

Mario Faria

>> Right.

Mario Faria

>> Yeah. I want to ask about who does CDO report to these days? And I want to see if you have ETR data and if they are in sync. But just before we go there, I just want to say, you mentioned a few companies like Nvidia and Apple. They're raking in all the money there is. So in S&P 500 is setting a new record almost every day. Now it's over 5,500, but it's only those top seven companies, if you look at S&P 493, they're not doing as well.

>> Right. Starting to broaden a little bit.

Mario Faria

>> Yeah.

>> Yeah, but you're right. It's the Nvidias, the Broadcoms, Microsoft, Google, Amazon-

>> Yeah, all hyperscalers-

>> Yeah....

>> Meta, Azure OpenAI.

>> It feels like it's a virtuous circle where all the CapEx money is going to Nvidia, and then it's coming back to the CapEx through companies like Anthropic, where the investments are made. And it's a little bit like 1999 in that sense because back then it was all advertising on portals like Yahoo and VC funding. But there are some similarities-

>> moment, the late '90s, all the companies and all the investment create an amazing network infrastructure and data centers infrastructures throughout the world.

>> It paid off. It paid off already.

>> Yeah. It took some time. You might remember lots of companies that we had seen in the mid-late '90s, they just disappeared.

>> Oh, yeah. Global Crossing, Enron, they were laying all this fiber.

>> Yeah. My favorite is one of the most spectacular failures of dotcom boom was Webvan, and Webvan's entire idea was to deliver groceries to your home. But we didn't have the fast network. We didn't have the social. We didn't have mobile. But today we have all that, and today we take it for granted.

>> You can make a business out of it.

>> So same thing's going to happen to GenAI. It's like some shine will come off, and then people will be like, "Oh, it bust. It went bust." But every single day, there's tremendous amount of development going on in AI, and it'll just keep getting better and better. And then maybe five years down the road, it'll become the de facto where we .

Mario Faria

>> That's right. And Dave, you mentioned something about risk and legal departments and so on, and with everything related to LLMs, and that you need to feed data in order for LLMs to work. And what happens if you release your data by mistake to another place that are going to be using your data where you don't have permission to do that? What will happen to multinational corporations where the data is stored in some places in Europe that you cannot export, and all of a sudden your model starts to working here in United States? So we'll see that new problems that we have never thought about will start to appear. So I believe that we're going to be going through a tremendous moment, new jobs that we even think about that the last five years are going to pop up. But at the same time, AI will bring more productivity that will allow some of the jobs out there to disappear. It's not going to be taking overnight, but it will happen.

>> All right. We got to leave it there.

>> No, I just want to close the loop on the CDO. Who does the CDO? Is it business, CEO? What are you seeing? Who's the CDO reporting to?

Mario Faria

>> The majority, I would say for the large companies, CDO has a large propensity to be working for a CIO.

>> CIO.

Mario Faria

>> Yes, for the large companies. But when you go to a smaller company, sometimes to a marketing operations to the CEO.

>> Yeah.

Mario Faria

>> Okay. So it really depends. So again, it depends is the answer.

>> Is that the same, ETR?

>> I don't have the ETR data on that. I don't know. It's all lumped together at C-suite. They just added that, but that's a good question to find out. Guys, thanks so much. . Pleasure having you.

Mario Faria

>> Thank you. So nice to see you again.

Mario Faria

>> Likewise.

>> For Sanjeev Mohan, Dave Vellante, we're here at CDOIQ 2024 in Cambridge, Massachusetts. Keep it right there.