Empowering AI Practitioners

[music]

0:00:04.1 Satyen Sangani: Welcome back to Data Radicals. In this riveting episode, I sit down with Wendy Turner-Williams, CEO of The Association.AI and former CDO of Tableau. Wendy discusses her mission to create a global community focused on ethical AI, highlighting the critical roles of AI and data practitioners in shaping the future. We also explore the unique challenges small and medium-sized businesses face in navigating complex regulations and the need for a cohesive data strategy that encompasses privacy, ethics, and security. Don't miss this engaging conversation filled with actionable advice and thought-provoking predictions from one of the industry's leading voices.

0:00:47.2 Producer: This podcast is brought to you by Alation, a platform that delivers trusted data. AI creators know you can't have trusted AI without trusted data. Today, our customers use Alation to build game-changing AI solutions that streamline productivity and improve the customer experience. Learn more about Alation at alation.com today on Data Radicals.

0:01:08.0 Satyen Sangani: Today on Data Radicals, we're joined by our old friend Wendy Turner-Williams, CEO of The Association.AI. Wendy has led digital transformations for over 20 years, shaping architecture, data, IT, and product teams at major tech companies, including Salesforce and Tableau. She also is the author of Unleashing the Power of Data with Trusted AI, a guide for board members and executives on using AI to drive growth and make smarter business decisions. Wendy is known for blending strategy with community building to inspire change and innovation. Wendy, welcome to Data Radicals again.

0:01:43.2 Wendy Turner-Williams: Yes, thank you so much for having me, Satyen. It's been a while since we've had a conversation, so I'm excited to catch up and get into this. And thank you for that wonderful introduction.

0:01:55.7 Satyen Sangani: Yeah, it's been too long and it's been fun to sort of watch you and your evolution since the last time we did this. And you may not know this, but yours was one of the most popular episodes that we've ever had so it's fun to bring you back on and continue the tradition of Wendy on the show. I guess the biggest and newest thing that you've been working on is the Associate Institute, and so why don't you start and tell us about what it is and why you started the organization.

Enabling the AI community

0:02:20.8 Wendy Turner-Williams: Yeah, so once I left the Tableau and Salesforce space, one of the things that I found myself really wanting to focus on is how do I support the data community at larger scale. And what was interesting about that is that, that was right in, I think October of 2022, right before AI really hit the major market in November, etcetera. And so once that happened, I decided to take that concept of how do I enable community, which we talked about at the beginning, at a much larger, even global scale. And how do I do it at a time where AI transformation is really at the forefront of what everyone is focusing on and struggling with as well. Right?

0:03:13.5 Wendy Turner-Williams: Not just focusing on, but struggling with. And how did I enable the people who work in the fields? If you think about the transformation, the people who work in the fields of AI, data, privacy, ethics, security, robotics. And robotics, I'll come back to in a second. They are the most important people in the world, and that's because they have their hands on the physical keyboards implementing the AI transformation.

0:03:46.7 Wendy Turner-Williams: If you think about what that means, they control the data, they control the access, they control the quality, they control the policies, they control the models. These are the people who matter. And so, for me when I think about an ethical and quality AI transformation for all, it's the people in those jobs who really resonate for me.

0:04:12.8 Wendy Turner-Williams: And so, I started the Association as a means to create a neutral, safe platform that enabled us to bring all of these different disciplines together, to basically create communities, drive conversations, crowdsource knowledge, and even focus on practitioner-driven technical implementation standards that bleed across our disciplines.

0:04:43.7 Wendy Turner-Williams: Back to, like Satyen, you've worked in data for a very, very long time. But if you think about some of these use cases coming up with AI, and especially if you think about some of the regulations that are coming, pretend there's going to be... Well, not pretend there will be a GDPR re next. In that GDPR re next, they will define new levels of personal configurations that you need to enable to support different AI model use cases where you, as an individual, can opt in or opt out.

0:05:18.2 Wendy Turner-Williams: An example might be, you have a regulation that says you, as a human, have the right to opt in or opt out of a diagnostic ML model for diseases. You opt in. Okay? Then you, as an individual, have the right to configure, how do you want that diagnostic. Can it come to you from an actual bot? Can it come in your email? Do you want an appointment? That level of granularity is coming.

0:05:47.7 Wendy Turner-Williams: And to do that, you have to implement things like data categories in addition to classifications. You have to have clear rules related to who owns that data and who doesn't. You have to have clarity in regards to your retention policies to support things like the right to be forgotten, etcetera, of those use cases.

0:06:14.5 Wendy Turner-Williams: So, it's not just like a cyber problem, it's not just a privacy problem, it's an end-to-end data management problem, where those are different dimensionalities of managing that information in order to meet that regulation. So back to the association. You can't take a siloed approach, right, to AI.

0:06:38.3 Wendy Turner-Williams: We have to bring these different disciplines that I talked about together, to have end-to-end conversations about our deep expertise and subject areas and the touchpoints, to really enable that trust and that ethics kind of scenarios. That's how the association came to be.

AI is interdisciplinary: Why a holistic perspective is needed

0:06:57.5 Satyen Sangani: And of course, there are other organizations, some of the major analyst firms like Gartner and Forrester, and they certainly have their coverage areas. And often, their coverage areas are exactly what you're discussing. They're a little bit more siloed because they have to focus and there's domain expertise.

0:07:14.5 Satyen Sangani: Is it that integration that was the white space? I mean, is it the fact that these things are not talking to each other, or people are not able to see sort of that integration well enough that caused you to start the organization?

0:07:26.8 Wendy Turner-Williams: I think there's two angles, actually. So, one is the integration, right? No one's looking at it from a holistic perspective. I'll give you an example. Regardless of Gartner, McKinsey, or etcetera, an example is the US Presidential AI Act. If you think about the AI Act and the recommendations that came out of it, it was very NIST heavy.

0:07:50.9 Wendy Turner-Williams: But how many data practitioners that you know actually use NIST as a framework? They don't. They use most of the data management maturity kind of frameworks that NIST is not necessarily known for. So, it took a very cyber angle. And that's a regulation. That's not even just the research firms, etcetera.

0:08:15.1 Wendy Turner-Williams: So, one is, yes, we have to have kind of holistic end-to-end processes. The other piece is that if you think about these research firms, they tend to engage with and target executives. And one of the things, when I first started the association, I had a lot of friends in the field and peers and people that when I talked about starting this group, were like, "Let's start this really elite high-end kind of group that's just for CDAOs and the best of the best."

0:08:42.9 Wendy Turner-Williams: And I'm like, how many of those exist, right? You think about the CDO club and you think about Ivanti through Gartner. You think about, you know, all these different groups. I'm like, executives, frankly, don't have their hands on the keyboard. They're not the ones on the front line actually doing the work.

0:09:04.5 Wendy Turner-Williams: And what happens is that when these publications come out, they tend to target executives in regards to the messaging and how do you kind of sell that this strategy is happening so you get executive buy-in or what's to come. I want to actually target the people who are literally writing the code for the models.

How "Right to be forgotten" exemplifies the need for practitioner tools

0:09:21.0 Wendy Turner-Williams: I want to target the people who are putting the controls in to actually monitor, against usage and protection, etcetera. I want to target the people who are building the ETL and the storage systems to ask the right questions about the usage, to make sure that it's the right decisioning, that the ethical lenses are being applied. And I want to arm them with technical implementation guidelines and requirements that actually meet regulations.

0:09:53.3 Wendy Turner-Williams: I'll give you an example of how this was a proven gap. Back to GDPR. Perfect example. There were a lot of publications that came from McKenzie or Gartner related to GDPR and what it would change in regards to technologies and processes, etcetera. But who came and actually filled the gap between the regulatory requirements and the tech vendors with, here's how what you actually do to implement various technologies in a framework approach that allows you to do configuration-based scalability to support any regulation going forward.

0:10:31.9 Wendy Turner-Williams: An example of this, back to data categories I mentioned earlier. To do something like the right to be forgotten, you not only have to understand if you own the data or not, you have to also be able to understand if that data is required for your business. So an example in software, I'm sure Alation went through this probably as well. In software, something like an IP address, right?

0:10:57.6 Wendy Turner-Williams: IP, per GDPR regulations, is owned by the customer. But you as a company, especially in a cloud service kind of environment, you use IP to authenticate that that's actually the person who has the license and that they get the right access to what they own. So, it's a threat detection, it's a security, it's a trust aspect.

0:11:21.6 Wendy Turner-Williams: And so, many tech companies defined that that was data that they need to be responsible use of their company and to operate their company. So, you need to be able to define at a field almost level, whether structured or unstructured data, what that data is and what it's used for. And then track the lineage through all of your systems so that you can truly either implement right to be forgotten or not. That requires things like categorizations. And there are ISO standards around these that people can use to have a common taxonomy across the board.

0:11:58.4 Wendy Turner-Williams: That taxonomy can be used very easily to start to configure those use cases I just talked about around like diagnostic ML models and, you know, is it email, opt-in, and preferences related to those things. This creates configuration-based, almost like regulatory compliance data handling standards that you can actually build into the data operations to treat these things in real time, versus a compliance control that's monitoring after the fact.

0:12:27.5 Wendy Turner-Williams: Once you have bad data in or handling in, it's kind of too late to fix it usually, right? There's so much more. So it's just ways to get more and more proactive. That level of guidance, I didn't see anything like that coming from Gartner or McKenzie. I didn't see anything happening that was targeting technical implementation, basic guidelines that a person a year out of college could take this and say, "Here's what I actually need to do," underneath the policy requirements that my policy team has actually put together. Like, here's all of the pieces that need to come together to actually support this.

0:13:06.8 Satyen Sangani: And this sort of practitioner-level knowledge is, I mean, to your point, a little bit of a black art. And it does somewhat depend on the technical stack that you've got. Because if you're running one stack versus another, how you do that work is going to be very different. And then it also depends largely on who you are and what your true problem is.

0:13:28.6 Satyen Sangani: Because then, of course, you also have this problem of implementation in the financial services industry might be very different versus implementing in the software industry, or if I'm a smaller firm versus a larger firm. And so, getting all that situational context to be able to get to a specific recommendation is where all the pain, to your point, lies. Pragmatically, how do you help your members navigate that? 

How does TheAssociation.AI put regulatory theory into practice?

0:13:49.1 Wendy Turner-Williams: Yeah, so what we've done is we've actually established a ethics and policy kind of board. And that board, again, you've been in data management for a very long time. We're following typical governance kind of protocols and frameworks. We've got an uber kind of policy and ethics board. We've got discipline-specific, so think cyber versus data versus privacy specific, almost like working groups who make recommendations based on what's happening with the regulations or with the risk, and the compliance needs.

0:14:24.9 Wendy Turner-Williams: And then we basically have actual industry groups. So the first group that we actually established was focused on what does ethical AI and healthcare mean? And how do you actually ensure that? So we've got, you know, practitioners, whether it's actually software providers, or it's actual hospital instances or freelance doctors, or the professionals groups like HIMSS, etcetera, all coming to the table to say, what does this look like? 

0:14:53.8 Wendy Turner-Williams: Then we can then break it down into further granularity. Meaning like, hey, this is what we would actually do as the baseline recommendations for say, a small company. Or a small doctor's office. Or here's what you would actually implement if you're a major software provider who's focused on DNA, like 23andMe. And what you should do.

0:15:14.5 Wendy Turner-Williams: There's a scalability, but you just have to set up the right systems of decisioning to feed into each other and actually give them each the space to operate based on what their actual focus says. The goal is again, there's not a one-size-fits-all, which is why the regulations are often not so clear. But also, it's because the regulators themselves just don't understand data. They don't understand data, they don't understand cyber, they don't understand cloud services. They don't understand.

0:15:48.3 Wendy Turner-Williams: Examples would be here in the US or anywhere else, AI is a global initiative that knows no borders. So each state or each country having their own AI policies or privacy policies, frankly, doesn't make any sense. [chuckle] Because most people, especially if we're on cloud, you may not even know where your data sits.

0:16:11.3 Wendy Turner-Williams: So again, there's basic principles and there's basic practices that you can define that are tech agnostic, that you can still have your own tech stack and your own tools, and there's lots of solutions and players that work in those components. But you can give basic kind of guidelines to say, here's the steps and the processes and the pieces that you need to put in place, and here's how they form together to create an encapsulation of trust.

0:16:41.0 Satyen Sangani: So as you've done this work, what if you learned... The regulations are changing, they're like literally real time. And you don't even know what's going to pass and what's not going to pass. And despite the fact that like AI is all the same thing, it does seem like in the US at least there's going to be a patchwork of state-level regulations that are going to get implemented starting with California.

0:17:05.2 Satyen Sangani: What have you learned, and what is the now that we're past sort of the initial waves of GDPR and CCPA, what is happening and what are the biggest challenges organizations are going to face in implementing and being compliant? 

The problem with AI regulations: Lack of clarity and over-complexity

0:17:17.2 Wendy Turner-Williams: Yeah. Well, I think the biggest challenge is, again, lack of clarity and over-complexity. A, you don't know exactly what you need to do and you don't know if you actually are required to do it or not. Because again, the regulations are so unclear. You may not know where you sit, etcetera.

0:17:39.3 Wendy Turner-Williams: But then in addition to that, I think that overcomplexity. There are too many regulations. At least, like in the United States, there needs to be a United States policy, that the various states need to implement. If you're a smaller type of company that's only within the United States, then you have a much clearer set of rules that you need to meet as far as some type of bar. I think that the complexity and the pace of change, like you said.

0:18:07.4 Wendy Turner-Williams: This is just starting. We're at the very tip of the regulations to come. Responsible use or ethics, even as a concept, means different things to different people, including regulators. I had a call with a State Department leader here in the State of Washington not that long ago and we were talking about the association. This is a gentleman who, by the way, is very tech-savvy and is a former CISO. Okay?

0:18:38.8 Wendy Turner-Williams: And when we talked about the association, one of the things he was super excited about was the fact about these standardized requirements. And what he was sharing is that it is so important for this level of information to come out because very few businesses become big, large enterprises. Most businesses are small to medium-sized businesses.

0:19:00.8 Wendy Turner-Williams: And the reality is that many of them, A, do not have the resourcing, who has the knowledge or experience around these things, to interpret them or implement them. Or in addition, they may not be following the regulations. Meaning, what's happening in the news? Or do I need to do this or do I not need to do this? 

0:19:21.2 Wendy Turner-Williams: But for him, he was talking about so much of the Attorney General's budget and spend goes towards suing small and medium-sized businesses who do not meet regulations. And they don't meet regulations because they don't have the budgets to hire experts. Or they may not even be following the trends and know that they have to. Okay? Know that they have to.

0:19:46.5 Wendy Turner-Williams: So this is where this whole, like, look, anyone can join the association. It's free... This is people who work in your same fields; you can learn from each other. Again, we need to kind of get rid of these silos that happen between the privacy team, and the cyber team, and the data team. They need to be working together. Because there are things, you know, they all have existing touchpoints. And this is a way that we can actually improve and uplift compliance, and again, trust and ethics across the board for any type of company.

0:20:19.2 Satyen Sangani: It's interesting. I would have not expected that small to medium-sized businesses are the ones that are getting attacked. Or not attacked, but at least, you know, driven to compliance. And this is, I think, one of the biggest problems with the regs, and to your point, the complexity.

0:20:35.2 Satyen Sangani: Which is that if you have such a complicated compliance regime that otherwise good people who are well-meaning and trying to do the right thing are turned into noncompliant individuals, or worse, criminals, you then lose credibility within the overall sort of regulatory regime. And so then, the entire system loses power. Makes total sense. And so, how's it going? 

Why AI practitioners need places to share knowledge

0:21:01.5 Wendy Turner-Williams: It's going really well. We have grown. We launched this last October and we've had like, I think, 350% growth month over month. We're global. We've launched chapters, not just in the US but we've got a chapter spinning up in London, we've got one going in Singapore, there's one that's spinning up now in Dublin.

0:21:22.0 Wendy Turner-Williams: Anyway, it's going all over the place. And again, it's interesting because people are looking for connection, which you kind of lose a little bit with AI. People are looking for shared thought processes, shared experiences. And with the shift around politics that has happened the last few years, where people are starting to move away from things like Facebook and other things because they don't like the sheer political aspects of it.

0:21:49.1 Wendy Turner-Williams: More and more professional networks are starting to get closer and closer in regards to blending kind of work with community and thoughts and building relationships that way. So it's been growing. We've gone from zero to over 13,000 members in a little over a year.

0:22:08.9 Satyen Sangani: Incredible.

0:22:09.8 Wendy Turner-Williams: Yeah. We're also working with universities. Again, I'm on the board at the University of Washington and of course I'm part of the Carnegie Mellon Chief Data and AI Certification Program. We're offering this to STEM students because they're the next generation. And not only do we want to have this knowledge share transition, but there's a constant ecosystem of knowledge share transformation that needs to happen with new people coming into the field.

0:22:36.0 Wendy Turner-Williams: And we can do more to support them to get a job and to get the education they need and to actually support the standards and the policies and get their voice, and is this work, does it not work. To even produce trends or heartbeats that are true reflections of the people who work in the fields, versus Gartner or McKinsey who a lot of times make up things [chuckle] based on what they want to have happen in the market, we can bring a voice to the actual practitioners and it can have a huge impact.

How has AI impacted data budgets and the CDO role?

0:23:08.4 Satyen Sangani: The notion of focusing on the practitioners is pretty timely, considering that there's a lot of discussion around the CDO where many of these practitioners might have lived. Some of them, not all of them. Some of them might have lived under the CISO, under the CPO. But today there's all these articles coming out.

0:23:25.4 Satyen Sangani: There's one series in Qstar about the Chief Data Officer and how they are under attack and that the roles are going away in some cases and that the data teams are often being merged under the CIO. One, I guess this is a trend that you see and believe in. I literally just had another interview with another analyst at another firm who said, "No, no, no. We see the CDO actually increasing in their capability." What are you seeing about the CEO?

0:23:51.6 Satyen Sangani: And given that you're, on some level, by dealing with a practitioner, you're sort of on some level impervious, I would imagine, to this trend. Like whether they live under the CDO or the CIO or the CPO or the CTO or the CXO, who cares? They still have the same problems to solve. But what are you seeing and what impact is it having on the work and the practitioner? And are you seeing data budgets compress, expand? Tell us about the lay of the land.

0:24:17.7 Wendy Turner-Williams: There's been a little bit of a shrinkage. I think the shrinkage basically came out of like the tech layoffs. I'm sure you saw within your own networks, and I know I did within mine, CEOs actually got hit pretty hard the last couple of years. The last three years. I think that that's a cycle and I think that that, again, I think data's always been really hard and the CDOs have not been very good at marketing themselves. We get so focused on doing our jobs, and our jobs are so complex.

0:24:49.8 Wendy Turner-Williams: And it is so much harder than I've got a metric on a report and so I know data because, yeah, I see a metric. There's so much more to it. And CDOs are not very good at explaining the complexity and really kind of spotlighting the impacts and the value that they have, based on actually collecting data about their own programs. [chuckle] And the impact.

0:25:16.8 Wendy Turner-Williams: So I think that that's been a natural kind of evolution that's happened along with the tech layoffs that caused a slight shrinkage. But I think that AI is going to shift that majorly to where we're going to see the combination of the CDO and CAI into one program.

0:25:33.7 Wendy Turner-Williams: AI is just a data producer and consumer. That's what it is. It eats data. And it will also eat your bad data. And all these years of technical debt and data debt that has happened within companies is going to come out with AI. Because people are just kind of pointing triggers right now and putting AI everywhere.

0:26:00.9 Wendy Turner-Williams: So things like hallucinations, or things like even copyright, or things like bias, these are actually data problems. They're data problems. And that's going to position, I think, CDOs in a different way because they're not just this underground IT kind of group who's just pumping out things so that you can basically, you know, get access to content on a report. There's an art to it.

0:26:28.4 Wendy Turner-Williams: And again, data management is extremely complex. And then you've got the analytics and you've got the data science and you've got architecture and you've got governance. All these things are their own disciplines and all are necessary to ensure AI is successful. So I think you're going to see more merging of the roles. I've seen a lot more upticks. I mean, you know, the federal government has required CDOs in every department, and that's been over the last couple of years. They're doing the same with CAIOs.

0:27:00.6 Wendy Turner-Williams: US is kind of leading this trend and there's been a dramatic uptick in the amount of CDOs. Anytime you have that uptick in those mandates, you're also going to see some shrinkage afterwards. Because not every CDO is the same either. And I don't think from company to company there's a clear clarity around what it means to be a CDO.

0:27:22.6 Satyen Sangani: Which is a phenomenal observation. What do you think the core skills are of a CDO? What do they need to learn and what are the, I guess, anti-patterns? 

0:27:31.5 Wendy Turner-Williams: Yeah. So to me, I think the core skills are you have to be a good business person, number one. I think a CDO sits right in the middle of business and technology. So you need to be a mix of both. And this goes back to again, there's different flavors, right? Some companies a CDO just owns analytics. At other companies the CDO only governs. At some companies a CDO maybe owns both of those but they don't own their own platforms to make sure that they're successful. That's the CIO and the IT group.

0:28:05.6 Wendy Turner-Williams: So depending on the company and the way that they're modeled, it looks slightly different, but I think you have to be an influencer. Because you're all about influencing without authority. You have to be able to navigate politics, which was our last conversation, because sadly, data is incredibly political because it's everywhere, everyone wants it, but no one claims it, especially when there's problems.

0:28:32.2 Wendy Turner-Williams: And as CDOs don't own the business strategy or the business applications, they get to collect the information, try to feed back, and normally they're an afterthought in that process. So they got to be able to navigate those political angles. They also need to be a mix of...

0:28:48.6 Wendy Turner-Williams: I mean, it's not about producing data anymore. It's about responsible use of the ethics. It's about data protection and loss prevention, or topics like quantum encryption. And how do we prevent hacking and how do we make sure that data is even more protected than ever, especially as we have AI coming in and its ability to hack.

0:29:16.9 Wendy Turner-Williams: So I think that we're going to see more of emergence of roles between the CDO, the Chief Privacy Officers and the CISOs. And I think that you're already starting to see that in a lot of the big tech companies, like Microsoft.

0:29:26.3 Satyen Sangani: I think it also depends on, at least what we're seeing is it really depends on the type of company you're coming from.

0:29:30.9 Wendy Turner-Williams: And the size.

0:29:32.6 Satyen Sangani: For sure. But there is a lot of ownership under the CIO. And it's an interesting question, like, how do you think the CIO is going to fare as they do this work? I mean, are they cut out for it or are you worried about that trend? Is that a trend that you're seeing?

0:29:45.5 Wendy Turner-Williams: Most companies don't have a complexity of multiple engineering teams. Like organizations. Right? So you normally have a CIO and the CIO is often responsible for data plus all of the infrastructure, etcetera. In my experience, I think of CIOs as being focused on containers. They are focused on what's the business strategy and what's the infrastructure we need in order to support that strategy and how do we operate it.

0:30:16.1 Wendy Turner-Williams: And when I think about a CDO is, it's like a bucket. Here's a bucket. But in that bucket's water, and that water is data. In my experience, what tends to happen is that the CIOs get overly focused on the infrastructure and the technology, versus what's the actual data needed out of those technologies in order for us to answer XYZ questions. And it's a different skill set, too as well.

0:30:51.7 Wendy Turner-Williams: So I think that again, I think that it's small, medium-sized companies. I think you're going to just see the evolution of a CIO role where people get more and more experience and a little bit more focused on the data aspects. And maybe they don't have a CDO. Or you have a CDO plus a CISO plus a CPO or whoever that report to the CIO. And I think that's fine. It just depends on your org structure and what you need. The point is I think that there are different discipline types.

0:31:18.8 Wendy Turner-Williams: But here's one thing that I mentioned, I think, that the evolution will be this merging of organizations around cyber and privacy and data. Because what is cyber? Cyber is about protection of the information. If you don't actually understand the information and the data and the actual risk associated with the data at a granularity level, then cyber's often not very data-driven in regards to their own risk management, their own protection.

0:31:50.3 Wendy Turner-Williams: And what's privacy? Privacy is about a dimension of data, a customer dimensionality of data. So I actually think that we'll eventually see some shifts that we have, CISOs and CPOs and maybe even CEOs, because the infrastructure, reporting to a CDO. And the CDO should be the CEO's best friend because what's their job? Their job is to enable cross-company strategic decisioning with high quality information that can be used by either humans or AI.

0:32:29.1 Wendy Turner-Williams: Protection of that is a grain around that area. Privacy is a grain around that area. CISOs and the containers, the containers should be chosen after you align the data strategy with the business strategy. And you have clarity around, what are the data gaps and the processes we need to plug. Then we put the technology. Okay?

0:32:56.4 Wendy Turner-Williams: And if you do that, there will be a lot more ROI in those investments, and you'll have a lot more trust in risk management in a more proactive way, back to the original actual SDLC process and the design. And that's the evolution that, for me, I'm like, it's already happening in big tech companies. Again, Microsoft has moved like their old data platform organization in Azure is now within the CISO's organization. Because they're seeing this.

0:33:29.9 Wendy Turner-Williams: Again, AI is forcing certain types of alignments. Some companies might do it under a CISO, some may do it under a CDO. Depends on that grade of the CDO, if they're technical and they're not, and they understand all these different aspects that I'm talking about. The same thing to a CISO.

0:33:46.0 Wendy Turner-Williams: Because frankly, CISOs get overly indexed on the compliance and the controls, and they often lock the data in a closet to where it's reducing agility and accessibility, or even discoverability of data, in a way that actually enables the company to be agile with the market. Or with other things.

0:34:09.4 Wendy Turner-Williams: So again, to me, if you do it through the CDO, you get that kind of cross-discipline. You get the focus on actually the strategy and the right questions, and then you put the technology and the protection and other things in the right lane based on the actual need.

The rise of the CIO: Risk talks

0:34:23.3 Satyen Sangani: Yeah, it's such an interesting conversation. I mean, there's all these competing dynamics. Another friend who happened to be a CIO, had observed that the CISO has gained so much power because they basically just have an objective job to basically keep information secure. And so the CISO, the CSO, is getting tons and tons of power.

0:34:49.3 Satyen Sangani: And therefore this person's observation was, now the CIO really wants to own AI, because broadly speaking, AI is the new app. And the thing that occurs to me is the baseline skill is what you said initially, which is, if you just can't talk to me about my business, and in business terms, if somebody just keeps on telling you about the problems, like if my head of marketing comes up to me like, "Well, you know, I just got to tell you about the leads, and the leads were just like, not really great. And we have this process and when we gather it, there's this field." I'm going to be like, "I don't care, just make your number."

0:35:23.9 Satyen Sangani: So, I do think that the core skill of just being able to be a business person and talk about outcomes, whether you're a CISO or a CIO or a CDO, invest there, because broadly speaking, otherwise this all becomes super gobbledygooky too.

0:35:36.7 Wendy Turner-Williams: Well, it goes back to a CEO, again, they tend to support the business teams and be involved with the business strategy. Because again, people think about the pure containers, right? As a part of that. The business applications, they think about it. So they'll bring the CIO to the table and the CIO can do a good job explaining the business processes and that this process ties to this process and through this application, etcetera.

0:36:03.0 Wendy Turner-Williams: The CISO's world is very, very clear. They're all about risk. And risk talks. You could be a CEO and not understand things like threat detection or things like how do you do a key ring process or king ceremony or all this other stuff. You don't need to know that. What you need to know is that you're not getting this SOX certification, or you could be fined XYZ by a breach, or that a breach will be on the news.

0:36:35.1 Wendy Turner-Williams: That's very, very clear. The data role and the AI role is not that clear when you talk about the problem sets or the monetization opportunities or the impact on the company overall or how you enabled like a strategy. Back to like we're not, CDOs and even CIOs, we're not very good marketers of ourselves. [chuckle]

0:37:01.3 Wendy Turner-Williams: We don't communicate very clearly. And many of us tend to ignore the actual business strategy and just focus on the day to day operations of the data. And that's never going to get you any type of real focus or invite in to the big table in regards to the strategic conversations.

0:37:24.1 Wendy Turner-Williams: But back to what I said, if you think about privacy and you think about the business applications and the technology as a tool that enables you to get decisions, with the decision being here's the questions we need to ask and here's the bars and levers we need to move. If you think about it in that way, it should be inversed.

0:37:47.1 Wendy Turner-Williams: There should be an inverse where data is actually more of a leader and a driver, and it would make the processes much more effective. But there's some upskilling and maybe some changes that need to happen in regards to the clarity about what it takes to be a CDO.

0:38:00.4 Satyen Sangani: This is the thing that sometimes you hear it in gender conversations, sometimes you hear it in ethics conversations, sometimes you hear it in data conversations, where people say like, "Well, you know, like I'm looking for a seat at the table." And often it's also the case that the seat is there.

0:38:16.8 Wendy Turner-Williams: Yes.

0:38:19.7 Satyen Sangani: But you have to also speak the language of the person heading the table and make sure that they know what benefit they get from including you there.

0:38:25.3 Wendy Turner-Williams: Right.

0:38:26.0 Satyen Sangani: And so it is very much the case that it's a two-way street. And so you're doing work, obviously, with Carnegie Mellon. You mentioned around sort of CDO training. Where's the curricula? Like how much is it helping? Is it changing and evolving? Are you seeing people recognize these issues?

0:38:39.2 Wendy Turner-Williams: Well, I think it's changing and evolving. In fact, this cohort this year, this is the first combined data and AI office. So they've made that stake that no, it's not just CDOs anymore, it's chief data and AI officers. Back to AI is a consumer of data and a producer of data, right? So, obviously the curriculum is evolving. I think that with AI and with the regulatory landscape as well, it's going to continue to evolve.

0:39:08.3 Wendy Turner-Williams: I mean, every day there's new things coming out from an AI perspective. It's going to do a lot in regards to data management and even things like the visualization. What's the point of a visualization anymore if you can basically just have your computer tell you what to do based on its interpretation?

0:39:25.8 Wendy Turner-Williams: A lot of the core kind of aspects of data and the different disciplines are going to radically change. As are with cyber and as are with other different things as well. [chuckle So I think that the curriculum is going to continue to evolve with the times, and I think the times are in the infancy for all of us in all disciplines right now in relation to what's coming.

AI predictions: What will change? What won't?

0:39:46.7 Satyen Sangani: Speaking of what's coming, that's a great bridge to my next question, which is, give me a couple of predictions. Next three to five years, what things are going to be really different and maybe what are the things that people are all prognosticating or forecasting that will not have changed at all?

0:40:05.5 Wendy Turner-Williams: Yeah. So I'm obsessed with this quantum encryption piece right now. I think that that is so important and I think that there needs to be a real tie between the CISOs and the CDOs to ensure that that's effective. Because again, it's very expensive to do. I don't think you have to do it for all data. This goes back to knowing what's in the data and what's really high risk and what's not. I think that that is something that is going to be really, really a big focus.

0:40:34.6 Wendy Turner-Williams: I think that there will be an integration of frameworks like the cybersecurity framework from NIST, with frameworks like CDAM, I think through, or DMM. Or all these different disciplines, I think are going to have to come together and start creating more holistic, kind of end to end scenario or even use case of data, type of frameworks based on what's coming in the types of granularity.

0:41:01.9 Wendy Turner-Williams: I definitely think we're heading towards, there's been a lot of talk about the agents and agent economy. I think there's going to see a lot of agent management functions. If you've got all these different agents and they're based on different scenarios and use cases, now think about that complexity. Again, back to the data. What are you exposing?

0:41:24.7 Wendy Turner-Williams: So I think there's going to be like a lot of focus on agent kind of management systems that come out to basically manage some of these things. I think we're going to see a lot more kind of configuration based compliance and controls, to where a lot of those controls are going to start backing up to the actual data creation processes.

0:41:46.8 Wendy Turner-Williams: I think we'll see more focus on data by design, so that if we do data by design, to align it to your business strategies, you can not only get those answers to those decisions that you need to make, but we can actually get to ethics, irresponsible use, we can get to security by design, we can get to privacy by design, all in those same steps.

0:42:10.2 Wendy Turner-Williams: And you can have intake processes that merge things like pen testing with the actual data deep dives, to go through those risk pieces in a different way. And I think we'll have all types of tools that come out on those particular pieces. RAG, of course is coming up. I think ESG is going to be a hot topic of course. Because AI needs a lot of power, right? [chuckle] Everyone's going nuclear for a lot of different things, all the big tech companies are building nuclear facilities.

0:42:40.6 Wendy Turner-Williams: But if you think about not just the compute, think about the networking, just bandwidth in itself. You're going to see telecommunications companies and stuff start to scale and hyperscale on things like that, because there's a lot of different aspects needed to actually ensure that that funnel is open.

0:42:58.3 Satyen Sangani: Any trends that you are taking the under on, or things that everybody's saying is going to happen that you are less optimistic or bullish on?

0:43:07.3 Wendy Turner-Williams: Well, give an example or throw a couple things out and I'll give you an opinion.

0:43:10.0 Satyen Sangani: I don't know...

0:43:10.0 Wendy Turner-Williams: But flip it back to you. [chuckle]

0:43:13.5 Satyen Sangani: Yeah, that's a great question. Don't make me answer my own questions.

[laughter]

0:43:18.7 Satyen Sangani: There's a whole bunch of people talking about whether or not AGI is going to happen sooner or later, that a lot of people are talking about how much data governance is going to be totally automated. There's people who are talking about whether or not there's going to be the need for traditional analytics and where VI set goes.

0:43:37.4 Wendy Turner-Williams: So, analytics. I think I already hit on that earlier. What's the point of traditional analytics layers when basically you can integrate this right into your business applications or even right off of your stores, and you can automate decision trees with AI to just do an action. The analytics is to derive some type of outcome or action to an outcome, and that can be automated without the analytics layer.

0:44:04.5 Wendy Turner-Williams: So, I think that, in fact this is a good one because this is one of the up-and-coming things. Intelligent applications, I think is going to be just huge. I think knowledge graphs are actually more impactful than just LLMs. So again, I think, so the combination of an LLM with a knowledge graph where you can see decision trees and you can see like the controls and the measures. That allows you to actually monitor what AI is doing and to tweak the decisioning based on what you want that to look like.

0:44:38.3 Wendy Turner-Williams: That, I think, is going to be a big thing over the next couple of years. I think that there's going to be a big downturn in SaaS. What's the point of SaaS systems? They're really about business applications to guide humans into the processes and the information needed in order to get some type of decision or outcome or state.

0:45:00.8 Wendy Turner-Williams: I think that you can basically, with AI, I think that agent can run directly off your stores and you can basically start to automate that entire business decision tree and processes, and it makes most SaaS irrelevant. Okay? And I think that that will be a big trend happening over many, many years. Most people aren't ready for that, but it will come.

0:45:27.0 Wendy Turner-Williams: Back to data governance. I've always thought it can be automated and it should be automated. Back to the regulations. Things on paper don't matter. It's just an audit checkbox. You have to automate the processes and you have to incorporate them into the actual data engineering processes. Whether it's the models, whether it's the actual, like ETL, whether it's... Whatever it is, those things can be automated. And I know you and I have had many discussions in the past around how I have automated that at several companies. So that's totally capable of doing that. [chuckle]

0:46:01.6 Wendy Turner-Williams: As far as AGI, I think it's coming soon. I think that it may be further along than people are willing to say it is. I think that there's already been a lot of big names, that a lot of, not just companies, but a lot of big labs who have shut things down because of fear around certain things.

0:46:20.0 Wendy Turner-Williams: Obviously, I started the association because I have a lot of concerns that the regulations are just not there to actually ensure the protections we need as humans or as consumers, based on the intelligence that's happening. Let me give you an example really quick.

0:46:43.2 Wendy Turner-Williams: I'm sure you saw there was a teenager who recently committed suicide based on his interactions with an AI bot that actually was targeted and created for children. The bot would not say it acknowledged that it was a bot and would basically reinforce back to this teenager that he should kill himself to be with her.

0:47:07.1 Satyen Sangani: That sounds like everybody's worst horrible nightmare.

0:47:10.7 Wendy Turner-Williams: And this goes back to, it's not about creating sexy technology. Everyone's trying to rush to that, and I get that there's major opportunity. There's responsibility that comes with AI. And back to the association and back to the people that work in the fields. You can do things like how much time is this person spending with this bot, and is it a safe level or not, to do recommendations.

0:47:43.3 Wendy Turner-Williams: You could put in controls to help with underage minors. You could definitely put in transparency aspects in regards to this is a bot. Or, you know, there's things that can be done that basically will help to ensure safe practices and need to ensure safe practices.

0:48:02.5 Wendy Turner-Williams: And I don't know how many deaths or negative things need to happen before people start to take that very seriously and the regulations start to catch up. But the regulations move slow, sadly. And to me, it's back to the people that work in these fields to have a lot more say within the companies in regards to, here's what can be done and here's what should be done. And it wouldn't slow our innovation. There are safety aspects that we need to incorporate, and a lot of them are data-focused and related.

0:48:39.7 Satyen Sangani: It's a really, I mean, and certainly, we wish, I certainly wish that we would have that just for basic social media, let alone AI. The business model of these companies is entirely antithetical to such controls because essentially they get paid based on engagement and the engagement drives more advertising and more clicks and more views, and all of it is predicated on that model. It'll be super interesting to see what the business models that AI end up producing and what incentives that come out of it.

0:49:10.2 Wendy Turner-Williams: You know, if it's something for children, there's certain protocols or certain things that must be there. Right?

0:49:16.3 Satyen Sangani: Yeah. Look, I would be of the mindset that that should already exist. And it's so vexing as a parent of kids to watch how freely they are able to sort of both circumvent existing controls and also just to be able to limit, you know, to even have any discussion around this, it's really super hard.

0:49:34.0 Wendy Turner-Williams: And as a parent, for sure. We already have this with existing social media. I think there'll be a lot more focus on those things, but I think it's going to take some time. But I think that that focus will be an evolution for data experts. Back to I think we'll see a big expansion of the role and focus. Because it requires the data in order to do those things, and it requires the quality data to do those things.

0:49:58.3 Satyen Sangani: Well, I love your enthusiasm and you basically took the over on everything, from AGI to data management. The optimism and the enthusiasm is palpable. So with that, I'm going to say goodbye until next time. But thank you, Wendy, for coming on the show. I think everybody, and the conversation proves it, will enjoy the discussion. But thank you.

0:50:18.8 Wendy Turner-Williams: Thank you again for having me. Everyone go join The Association.AI. And it's free. We're building a community. Come build it with me.

[music]

0:50:30.3 Satyen Sangani: It's always great to have Wendy on the podcast. It's clear that she brings invaluable insights into the evolving landscape of AI and data management. From her extensive experience leading digital transformations at major tech companies, to founding The Association.AI, Wendy emphasizes the critical need for a holistic approach to data ethics and community-driven standards. She also underlines the importance of cross-disciplinary collaboration and ethical considerations in fostering responsible AI practices.

0:51:01.7 Satyen Sangani: Thanks to Wendy's passion and expertise, we see a compelling vision for a more integrated and proactive approach to AI and data. I encourage everyone to join her on this transformative journey.

0:51:11.0 Satyen Sangani: I'm Satyen Sangani, CEO of Alation. Data Radicals, keep learning and sharing. Until next time.

0:51:17.9 Producer: This podcast is brought to you by Alation. Your boss may be AI-ready, but is your data? Learn how to prepare your data for a range of AI use cases. This white paper will show you how to build an AI success strategy and avoid common pitfalls. Visit alation.com/AI-ready. That's, alation.com/AI-ready.