A Data Leader’s Guide to AI Success

Table of Contents
Introduction and Panelist Profiles

• 01. How should data leaders manage AI’s expanding impact in the enterprise? 

• 02. How does AI change the data leader’s role? 

• 03. How can data leaders serve the AI needs of multiple C-suite members?

• 04. What are some best practices to being a data leader in the AI era?

Introduction

A Data Leader’s Guide to Succeeding with AI in the Enterprise
Facing a very real “AI Trust Gap” that makes tech and business leaders hesitant about embracing generative AI as well as a fierce desire to drive competitive advantage using it, data leaders find themselves striking a tough balance. They are finally receiving the full attention of the C-Suite and their boards, but the requests being made and expectations being put upon them are often out of line with reality. What should a data leader be doing to succeed with AI in the enterprise? Data Leadership Collaborative gathered five data leaders from across financial services, technology, medical products, and a generative AI startup to discuss issues such as AI’s role in the enterprise, serving the needs of multiple leaders, and the skills required to productively bring AI into today’s organizations.

Panelists

• Aisha Quaintance -Vice President Strategic Development, RelationalAI
• Chitrang Dave -Senior Director, Enterprise Data and Analytics, Edwards Lifesciences 
• Deep Srivastav - SVP, Head of AI and Digital Transformation, Franklin Templeton
• Kevin O’Callaghan - Head of Data Analytics, Teamwork
• Solomon Kahn - Founder and CEO, Delivery Layer
   

01. How should data leaders manage AI’s expanding impact in the enterprise?

To set the tone for the conversation, a panelist read the recent 22-word statement on AI from the Center for AI Safety: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.”

Data Leadership Collaborative (DLC): Now that we’ve heard one sentiment about the potential societal impact of AI, it would be interesting to hear where each of you are with enterprise AI right now and how it’s being received in your organizations? 

Deep Srivastav: I tend to see AI with two different lenses. One is more tactical. You need to look at tasks such as business processes that AI can help us scale and automate as well as tasks where using AI would be impossible or not very cost-effective. As data leaders we have the luxury of getting a pretty clear enterprise-wide view of what’s going on and knowing what those data levers are. The other lens is more strategic and cultural and requires conversations with the CEO and business leadership level, as we explore possibilities beyond what we can do with human intelligence alone.

Chitrang Dave: I feel like there’s so much energy around AI right now. I often think back to the emergence of the internet or the iPhone. I remember when iPhone was released: all the executives in my company were walking around with Blackberries. Then everyone else started coming in with iPhones and this huge shift happened. I think we’re at that point with AI.   

Kevin O’Callaghan:  I’ve found with AI that the possibilities and potential rewards are limitless, but the asks are probably limitless as well. We’re looking at AI across the board here, not just in terms of what can we do to differentiate ourselves further in the marketplace by enhancing our product, but internally in terms of efficiencies, organization, and structure. We have to define the skills and build the knowledge to do that in both areas and be aware of the risks if we don’t. I’m sure plenty of others out there are seeing competitors adding AI and GPT and feeling that they have to get at it and do this, but at the same time we know if we don’t do it right we’re going to end up making a bigger mess down the line.  

Deep Srivastav: I would echo a lot of what Kevin said, again at two levels. First, we should be thinking about AI regulation, security, and also the privacy issue. How are we making sure no data is getting past our walls – or if our systems are leveraging any of the large language models, even if that’s hidden, is there an interaction happening with them? What are those models actually capturing and storing versus what we are? When there’s an interaction happening, you have to be super careful about it. And then very closely tied in with that is also the challenge about hallucination, right? Accuracy and sourcing of information are critical. You must take information back to its sources and decide how to catalog it.

Having said that, it’s very important to realize that AI is one of those once-in-a-lifetime opportunities. So how do we break the traditional paradigms so that we’re not thinking too small or using the traditional program management constructs to bring about change? My own mental timeframe has shifted. When somebody says two months ago in AI, that feels like a long time back because so much has happened in the past two months. We need to build the understanding in the entire organization that every day and every week matters in how we are thinking about AI. We almost have to think of a stochastic strategy for AI because it’s all on top of a constantly shifting landscape.

Aisha Quaintance: One of the ideas that’s been helpful to me is to flip the question back to the person who’s asking, “How are we going to do this with AI?” I would prefer to say, Let’s figure this out together. How do you want to do this? What are you hoping to get from AI? And then having them help create the solution, or not. You’re leading the witness a bit to show them that it’s not as easy as they might have read in the latest blog, but also bringing them in as part of the solution.  

DLC: It sounds like there may be an expectation that you’ve got enterprise AI mastered simply by virtue of your title.

Aisha Quaintance: Absolutely. You’re kind of supposed to know as the data leader what your history is with the new technology, whatever it is, and all the benefits and rewards. But the bottom line is we just kind of all got hit. We got thrown a curveball by understanding some of what this can do for us but not knowing what we don’t know in terms of how it’s going to completely affect an organization. That’s why I think it’s okay to say that and then to bring leadership into the discovery and qualification of what we’re trying to do. What do we think it will do to benefit the business and what are the real risks? And are we willing to take them together? That would be my advice because otherwise you’re going to guess and take a stab and possibly fall flat on your face because it’s such new technology. But we’re not supposed to be the experts. This happened to all of us, this didn’t just happen to the data leaders.

02. How does AI change the data leader’s role?

DLC: From what you’ve been saying, it sounds like AI is powerful enough as a force to shift your roles. What does that look like?

Chitrang Dave: I think one of the roles that we as data leaders need to play is that of an evangelist. That means we should really understand the technology, know what the sweet spots are, and where we can apply it. I would start with the evangelizing portion because that’s what we need to do with AI. Obviously, you need to be able to be clear-eyed about the potential but also be able to come down to the use cases. To get there you need to go deep into a domain and ask yourself: What are the functions? What are the tasks? What are the jobs to be done? We also should be taking a view that challenges some of that because I believe we’re at a point where we should be thinking about jobs not just as somethings humans can be doing, but a new kind of job we could be doing if we didn’t have some of the human constraints we currently do. 

Solomon Kahn: Jobs to be done is a good comment because most business leaders have a sense of the key jobs being done inside their organization. The question then becomes, what are the good candidates where AI could transform that job in some way? Some organizations will have a lot of those that will have high impact based on their domain and some won’t have that many. I think it’s important for any executive, data or not, to understand where they are on that spectrum and then plan accordingly.

I also think it’s fair to say that if you don’t have someone who’s able to steward the organization through this new transformation, you may not have the right leader. You don’t need to know all the answers for all of the specifics of how to implement AI, but sometimes in any profession you end up in a place where you’re put in a leadership role for something really important that’s happening in the world. It might not have been important a while ago, but now it’s important and you need to very quickly get up to speed on the background of what’s changed that falls under your domain. If your organization can’t rely on you to shepherd your organization through this, it’s a valid question for an organization to say, are you the right leader?  

Aisha Quaintance: Very well said.

03. How can data leaders serve the AI needs of multiple leaders?

DLC: That’s a good segue into our next topic: serving the needs of leaders who might have different agendas around AI. Should you be pulling these leaders into your meetings or pushing yourselves into theirs?

Solomon Kahn: My personal opinion is that the changes that are happening with AI are too big to only funnel through the data team and are too profound to wait for governance and other data infrastructure projects to be ready before you start implementing more widespread. You might not be able to build your own LLM until certain things happen. You might rely on third parties or off-the-shelf tools. But business leaders who implement AI vendor solutions based on what those salespeople are telling them and not their own expert data people are going to be in a lot of trouble. 

I’ve heard of enforcing blanket bands on ChatGPT, but then there are a thousand other popups and plugins and startups you won’t be able to keep up with using blanket bans. It’s just not possible. So you have to be more engaging and proactive with the folks who want to try this stuff out. I’m a startup founder now, but if I was leading a team today l’d be thinking about it much more holistically versus simply what are the engineering things that I’m going to be personally responsible for on my team? There is no way to contain the business from moving, so it’s better to be a part of that than to try to be a silo outside of it.

Chitrang Dave: I am of the same opinion that AI is too big, too profound, and too disruptive to be just part of the data group. Highly valued, highly trusted data leader partners are invaluable. That’s why I think Deep has a great role. He’s responsible for digital transformation, so he is at the right level, engaging with executives to build that trust. But I think these roles are few and far between. 

Kevin O’Callaghan: If you asked me six months ago when I was doing this, I would have said, we’ll focus on our governance and lineage. While we do that, we’ll look at the most pressing issues when it comes to data, identify which to fix and we won’t boil the ocean. Now, it’s how best can we quickly implement AI and one option may be to expand a data governance committee to also look at AI governance. That means decision makers have transparency,  have control over the priorities we’re addressing and their managers also know where the real pain points were. So for me it is a little bit more organizing and trying to get everyone else around the table with us in terms of not just saying “Here’s all the cool stuff,” but actually identifying the things we need to focus on first and go from there.

DLC: Chitrang made a good point about your role, Deep, as head of digital transformation. How do your leaders see that you have the right perspective on these issues and can collaborate with them to make the right recommendations?

Deep Srivastav: I feel fortunate that we already were on our journey before AI became a buzzword. We had a clear conviction that AI was the right direction to go, so things like ChatGPT actually helped to catalyze a lot of that. And we had some good success coming up with AI-native solutions for our clients even earlier. Now there is an enterprise-wide excitement around what we can do there. AI has shifted from a demand problem to a supply problem. Until now, my prior role was much more focused on how do you get people engaged on AI in both enterprise and with clients and customers. And now it is, how can we execute on it?

You don’t have to sell people on the potential. What you do have to sell people on is how do we go about it? What is a short-term strategy? What is a long-term strategy? How do we place our bets along those lines? So that’s where we see the difference. People come to us to become more knowledgeable in the space, but our ability to create a group of proof-of-concept initiatives that are happening across the board make us the connecting group. My biggest focus now is how do you do the blocking and tackling because so much potential can itself become very paralyzing and you can very quickly get lost in what can happen in the future. Or you can get lost in trying to execute a very tactical solution. And keeping a balance between the two and taking the enterprise along has been the more exciting part of the last few months.

Aisha Quaintance: One thing I’ve noticed in forums is that everybody’s insecure about a CDO-level role and doesn’t want to mess it up. There’s a little frustration or even imposter syndrome that people think they’re supposed to know. How do you give somebody that is newer into the role the confidence to say that they can lead the discussion on the data strategy but they don’t have to own it alone and come up with all the answers?

Deep Srivastav: If you are actually in the AI community, working with the scientists of the tech companies, you see a lot of humility. People know that it is not one person or one company that created this. There’s a lot of understanding that none of us have all the answers. And if you have the right people, then you can really create that dialogue of co-creation and collaboration. Gen AI differs from traditional AI because traditionally AI was forced to be in a black box. You had to know models and coding in a certain way. Gen AI is a lubricant for the whole industry. It allows people to have a much better dialogue than they could have. 

Aisha Quaintance: That is such a great takeaway, and it gets at a piece of advice I’d give someone. I’d say, now is your time. Don’t feel frustrated and don’t fall for the newest trending software or the new hype. Instead, how can you make AI everyone’s problem to solve together at your company? Because now you can bring key stakeholders and business owners into the conversation in a way that maybe you couldn’t in the past.

04. What are some best practices to being a data leader in the AI era?

DLC: That’s another great segue, because we can open up the conversation about the cultural and change management best practices for leading in an AI era.

Kevin O’Callaghan: That’s my world right now. There’s so many scenarios and solutions and I’ve had conversations with leaders looking to bring in a new solution and solve problems with AI. I counter that question by asking, What are we actually trying to achieve, and is a chatbot or LLM or other the right solution? And let’s give this a little bit of thought. I think it’s important to articulate a position. Then we can collectively look to organize and have a broad discussion across the organization in terms of what those goals are. Some will want to move on AI just for the sake of doing it. Others will want to move for the sake of actually getting a result out of it. That’s where the culture and change gets the opportunity to embed itself.  

Aisha Quaintance: Maybe it’s all the insecurity in Silicon Valley right now with everybody moving in and out of different roles. What I’ve seen happen lately is VCs or CEOs say that maybe they don’t have the right data leader in place if they don’t know how to answer that question about how we would implement a chatbot tomorrow. And how can you get in front of that, maybe by explaining that you’re part of these different communities who are looking at all these toys of the day that can plug into ChatGPT, and you wanted to bring it up proactively to discuss how we might want to implement this. Knowing what every last new thing is will be what gets the company into trouble. It’s the stewardship and strategic thinking of leaders that will actually be helpful. And it reminds me a bit of how it used to be in Marketing. Do we have the latest, greatest social media buzz person, and if not does the CMO need to go? When I was an executive recruiter I would say, Well, has anybody told the CMO though that you want to go in this direction? You just have to go there together.

DLC: How are you all thinking about workforce planning when it comes to AI, in addition to all the change management that has to go on? 

Deep Srivastav: I would say more than a culture of AI, this is just another leg in the rapidly changing technology landscape. I think it’s more being able to thrive in an uncertain environment that is way more fundamental than AI. If history is any indicator, there are always new opportunities. But how do you quickly move to the next opportunity, both at an individual level and as an organization? If you only make this a zero-sum game, then it doesn’t go too far. But if you create a significantly expanded opportunity set, then I think the question is much more about how do you translate or gravitate as the landscape changes. How fast you can move and evolve is much more important than what particular issue or what particular thing is getting automated today.

Solomon Kahn: Yeah, my future hires are not going to be any different than who my hires would be. I’m actually less bullish on AI for coding and replacing programmers. I’ve seen it, and I think that the challenge in programming is understanding where things fit inside complex systems. It’s not writing very simple scripts. I think you will never be able to get an AI system, or at least based on the technology that’s available right now, that understands very complex systems and how to program within those. Same thing with marketing strategies. I think AI can come up with ten ideas for headlines, but I think it takes someone with an expertise in the industry to look at those headlines and decide what you’re going to do with them. I think it’s going to empower people to be able to do more, which is going to be great. But I don’t see a substantially different number of people that my organization would need to hire.

Chitrang Dave: On the question of that coder you might hire, I would say they have to get the code, but more importantly can the person think mathematically? You still need the background, but you need it in a slightly higher order than what you needed in the past. There will be jobs that will be entirely eliminated, but at the same time there are going to be a lot of other opportunities coming. I believe everybody should be thinking of their jobs in terms of, “If I had five or 10 or 100 more people, how would I be doing it differently?”

Solomon Kahn: Can we go back to that first 22-word statement? Does anybody think that AI is a serious risk and that regulations should be mostly worried about the extinction of humanity? Or is that a deflection of what should be the more realistic risks and potential regulations that will lead to disruption of human life, but not extinction of human life? Which is my point of view.

Chitrang Dave: There are risks. I definitely see that. I’ve spent a career in life sciences and MedTech, and I like that approach of governance, regulation, and innovation within some sort of a framework. So I do appreciate that. And I’m very much in the camp of the Andreesen Horowitz article on why AI will save the world. Yes, there is real risk. Yes, we need to manage AI as data leaders. But the potential is huge. 

DLC: To stay in the Andreesen Horowitz vein, now that software has eaten the world, maybe AI will save the world.