Latest / Full Tech Ahead / Why CEOs are Failing Their AI Strategy

Transcript

Why CEOs are Failing Their AI Strategy · Full Tech Ahead

• 0:19
  SPEAKER_01: Hello and welcome to Full Tech Ahead.
• 0:22
  SPEAKER_01: I'm your host, Amanda Razzani.
• 0:24
  SPEAKER_01: And with me today, I have Kurt Mummel.
• 0:26
  SPEAKER_01: He is the head of AI strategy at DataIQ.
• 0:29
  SPEAKER_01: How are you doing today, Kurt?
• 0:31
  SPEAKER_00: I'm doing great.
• 0:32
  SPEAKER_00: It's great to be with you, Amanda.
• 0:33
  SPEAKER_00: Thanks so much for the opportunity.
• 0:35
  SPEAKER_01: Happy to have you on the show.
• 0:36
  SPEAKER_01: Can you share with our audience about DataIQ what services do you provide?
• 0:41
  SPEAKER_00: Yeah.
• 0:42
  SPEAKER_00: So uh DataIQ, we're the platform for AI success.
• 0:45
  SPEAKER_00: And what it is, is a platform targeting really enterprise customers, mostly global 2000 large organizations that uh want to be able to work on top of their data, wherever that data may sit, to build analytics, models, and increasingly AI agents using the best LLMs out there to transform their business.
• 1:05
  SPEAKER_00: So, you know, as a company, we've uh we've been around, kind of came out of the uh the uh the machine learning space over the past uh decade or so, but really now focusing a lot on accelerating the ability for organizations to you know take some of those incredible LLMs that all their favorite providers are offering them, combine those with the uh the enterprise data that they uh uh that they have, which is uniquely theirs, and turn it into yeah, agents that uh that run their business uh under the, of course, careful supervision of their uh of their human experts.
• 1:36
  SPEAKER_01: Wonderful.
• 1:37
  SPEAKER_01: Well, I believe Data IQ recently conducted a survey of about 900 or so CEOs.
• 1:44
  SPEAKER_01: Um, can you share what this survey was about and what information did you gain from it?
• 1:51
  SPEAKER_00: Yeah, yeah.
• 1:52
  SPEAKER_00: So this is the uh CEO confessions uh report uh 2026 edition.
• 1:58
  SPEAKER_00: Uh and uh I I really love this format.
• 2:00
  SPEAKER_00: It's a format that we've run a couple of times over the years, targeting different roles at enterprises, typically large enterprises.
• 2:08
  SPEAKER_00: We've done it for CIOs, uh, we've done it with CDOs, and now we're doing it with CEOs.
• 2:13
  SPEAKER_00: And the way that we frame it uh is the confessions.
• 2:17
  SPEAKER_00: What are the things that they would admit to uh if they're anonymous, if their, you know, their answers are not going to be attached back to them because it's only gonna be reported in uh in aggregate.
• 2:28
  SPEAKER_00: And we get these really uh interesting insights into what it's like to be in their shoes in the um uh, you know, well, the the current fast-moving uh space that we're all in right now.
• 2:41
  SPEAKER_00: And so here with the CEO confessions report, um, we really focused on the way in which the CEOs themselves are feeling this strain between the push to deliver with AI against then the uh the tension about controlling it, about controlling their spending on it, about uh being able to show that they're they're actually getting value out of it.
• 3:02
  SPEAKER_00: And then also, interestingly, where the decisions are being made uh in the company.
• 3:08
  SPEAKER_00: And it starts to reveal some really interesting tensions.
• 3:10
  SPEAKER_00: So, for example, one of the things that uh that I was surprised to read was that 70% of the CEOs that we survey um say that they own the AI strategy.
• 3:19
  SPEAKER_00: You know, that sounds, you know, that I guess that in and of itself is not uh is not shocking.
• 3:23
  SPEAKER_00: Um, but then only 60% are saying that they participate in a lot of or most of the AI related decisions.
• 3:30
  SPEAKER_00: And so there's this uh growing disconnect between ownership of the uh of the initiative and then the the actual decision making, which is going on to uh well to to deliver on the promises that then the CEOs need to go and defend uh in front of their boards of directors.
• 3:47
  SPEAKER_00: So, yeah, it's uh it's been a really rich and um uh valuable resource as uh I've been reading it as we've been talking to our uh to our customers about it.
• 3:57
  SPEAKER_01: That's interesting.
• 3:58
  SPEAKER_01: So they're not necessarily making this decision about the AI implementation, but then they're having to own that project and make sure that everything goes smoothly.
• 4:09
  SPEAKER_00: Yeah, that's right.
• 4:10
  SPEAKER_00: And you know, what what we're seeing then is that of course uh a lot of the core decision making is then being uh being delegated further down the uh the chain.
• 4:18
  SPEAKER_00: So within the executive suite, we're seeing that that's going to the CIO and then um you know a a head of AI, uh often underneath the CIO.
• 4:28
  SPEAKER_00: Um, but it's a real challenge for the CEO, uh, that chief executive officer, um, because again, they're the one who needs to be the face to the board of directors, uh reporting on that that initiative, but also the the public-faced and the uh the investor-facing face person who is uh who then needs to go and explain um the way this initiative is going.
• 4:49
  SPEAKER_00: And so this this growing challenge between um ownership and accountability, we think is uh is going to drive some perhaps some some rethinking of the way that uh that organizations set the objectives, set the measures for these different executive roles.
• 5:04
  SPEAKER_01: Yeah, absolutely.
• 5:05
  SPEAKER_01: It seems that maybe there's a lack of communication there.
• 5:08
  SPEAKER_01: I mean, from your experience, how could companies be doing this better so that everyone feels fully in the loop?
• 5:16
  SPEAKER_00: Yeah, so I think that um uh that the the blueprint is not written yet.
• 5:20
  SPEAKER_00: Uh, you know, we're we're in the middle of this uh incredible technological, economic, sociological change that uh that's being driven by the extremely rapid advance of AI.
• 5:33
  SPEAKER_00: And you know, I think you and I we're experiencing this in our in our personal lives, where we're discovering new ways to use AI, where AI in all of its different forms is uh uh entering our daily lives, sometimes entering our our relationships in ways that uh that are driving subtle changes.
• 5:50
  SPEAKER_00: Um and this is of course uh only exasperate exacerbated or you know made more significant within the enterprise, where these questions of ownership, process, et cetera, are really being uh defined for the first time.
• 6:05
  SPEAKER_00: What we're seeing, though, from organizations that are doing this well is that they are taking, you know, a kind of a learn fast and experiment quickly approach to it.
• 6:15
  SPEAKER_00: It is involving a lot of um, let's say, hands-on leadership from top executives, from that CEO and from that CIO.
• 6:24
  SPEAKER_00: Um, one of the things that we've observed is that uh that organizations where the the CEO themselves is, as an individual, a user of AI.
• 6:32
  SPEAKER_00: Often we see that uh that that is um translating into uh better directives, better understanding of the limitations and the potential of the technology.
• 6:41
  SPEAKER_00: You know, it's a bit as if in you know 1995, 1998, uh you had a CEO who uh still was not using email but was pushing a uh you know a very online strategy for the uh for the organization.
• 6:53
  SPEAKER_00: You might have some doubts.
• 6:54
  SPEAKER_00: Um so I think that it's really uh an opportunity for for organizations to make sure that everybody is digging in um and that they're testing the best approaches to uh uh to defining responsibilities, um, which includes then the responsibility for actually testing the technologies, building out new capabilities, and rolling them out to the business.
• 7:16
  unknown: Yeah.
• 7:16
  SPEAKER_01: And at the same time, I think another struggle that we hear about lately that I'm sure CEOs are dealing with as well, on top of these actual planned AI initiatives, you also have company employees using their own AI tools or unapproved AI tools uh throughout different departments and being able to keep track of that from a safety and security uh measure.
• 7:40
  SPEAKER_01: So, what advice do you have in that area?
• 7:43
  SPEAKER_00: Yeah, yeah.
• 7:44
  SPEAKER_00: And this is uh, I think really um interesting and really, in a way, unique to uh to AI because the technology itself, especially when it's presented as a as a chatbot, is just about as easy as any technology is to use.
• 7:57
  SPEAKER_00: Um, we're all using it uh in our personal lives.
• 8:01
  SPEAKER_00: Um, we we find it useful there.
• 8:03
  SPEAKER_00: And so it's understandable that the employees of an organization would want to use it within uh within their day-to-day work.
• 8:09
  SPEAKER_00: And if I were a CEO who were uh responsible for driving an AI-led change in my organization, I would be encouraged to hear that my uh my employees are wanting to use this technology.
• 8:21
  SPEAKER_00: Now, at the same time, there's real concerns about uh about the security, about the privacy, um, and about the cost.
• 8:27
  SPEAKER_00: Uh and especially, of course, if uh individuals are you know taking corporate documents, corporate data, and uploading it to the chatbot of their uh uh of their choice outside of that corporate firewall.
• 8:40
  SPEAKER_00: That, of course, is just uh is just not possible, right?
• 8:43
  SPEAKER_00: We can't uh we can't allow for that.
• 8:45
  SPEAKER_00: No organization can allow for that.
• 8:47
  SPEAKER_00: And what this drives is the need for these organizations to go fast, to be able to deliver that capability to their employees quickly in a way that is secure.
• 8:57
  SPEAKER_00: But then importantly, to uh you know, to go beyond that initial internal chat bot, internal copilot uh sort of initiative to actually start building out some of the more advanced AI capabilities.
• 9:08
  SPEAKER_00: Here I'm thinking about agents that are actually running parts of uh parts of certain workflows, running parts of uh certain decision processes within uh within an enterprise, so that they can not only address the immediate demand for, you know, just the assistance that uh you know that a chatbot can provide, summarize a document, reword something, but then also show their employees that AI can do so much more for the business than just again be a supercharged uh text processor for them.
• 9:37
  SPEAKER_01: Right, absolutely.
• 9:39
  SPEAKER_01: Well, so when we look at how quickly AI is advancing, all companies are trying to uh jump on how they can use AI most effectively.
• 9:49
  SPEAKER_01: But we're still seeing a lot of vendor decisions that are leaving employees unhappy.
• 9:56
  SPEAKER_01: They make these decisions and then when they roll them out, they're second guessing them.
• 10:00
  SPEAKER_01: Um, they're not immediately necessarily seeing that ROI.
• 10:04
  SPEAKER_01: So um, where are things going wrong here?
• 10:06
  SPEAKER_01: What's the bottleneck and how do you advise business leaders handle that?
• 10:11
  SPEAKER_00: Yeah, it's it's hard, right?
• 10:14
  SPEAKER_00: Because uh again, there is this pent-up demand.
• 10:16
  SPEAKER_00: There's a pressure coming from the board to uh to do something, to, you know, to basically to spend money on this and solve uh, you know, solve some problems with AI.
• 10:24
  SPEAKER_00: And what was, I think, one of the most shocking numbers uh to me and the report was that 76% of CEOs said that they regretted one of the choices, one of the vendor choices that they had made, um, and were feeling overly dependent on too few AI vendors.
• 10:40
  SPEAKER_00: And, you know, if we think about the events um, you know, of June and July, where certain models became unavailable due to export controls, uh government decisions, I think that throughout the uh the entire market, from the AI creators, the the developers of the models themselves, to of course the enterprises that are becoming increasingly dependent on those tools, that was really a moment that drove home the need to maintain, let's call it, strategic independence, flexibility among AI vendors, um, in the same way that organizations have historically um made sure that they've not had all of their eggs in one basket for other critical aspects of their operations, because it seems like you know, AI is increasingly becoming a geopolitical topic, um, which means that various governments around the world may start uh uh imposing restrictions on either the availability or the use or the performance of certain AI models, which means that it's critically important uh for enterprises to be able to choose a model, you know, work with a good model, um, but then be ready to switch, to be able to test and switch quickly, which uh which is one of the things that actually Did IQ really focuses on in uh uh in our product is making sure that there's that, let's say, independence from uh from any one of the AI vendors, um, in the same way that uh we've done that with uh with clouds and data platforms in the past.
• 12:05
  SPEAKER_00: So I think that this is a uh a really critical moment for organizations to yes, move quickly on AI, uh, yes, make uh uh make choices.
• 12:13
  SPEAKER_00: It's necessary.
• 12:14
  SPEAKER_00: Um, but at the same time, to make sure that you're maintaining a degree of strategic independence uh from any one vendor to make sure that you have that flexibility going forward.
• 12:25
  SPEAKER_01: Yeah, absolutely, because there is always that fear.
• 12:27
  SPEAKER_01: Um, we've seen, you know, a few AI tools that have become banned essentially for one reason or another.
• 12:35
  SPEAKER_01: And if you're a company that that implements these certain AI tools, that's a lot of money wasted if you're having to roll those back.
• 12:45
  SPEAKER_00: Yeah, yeah, it sure is.
• 12:46
  SPEAKER_00: And uh, you know, it's better to be able to certainly anticipate that uh that need to be able to switch um and make sure that you're not too tightly coupling any one of these things that you're building to the underlying model and that you can be able and that you can move relatively easily from uh from one model to uh to the next.
• 13:04
  SPEAKER_01: And also I know a big issue right now is token usage.
• 13:09
  SPEAKER_01: You know, companies have come to rely on being able to produce a certain amount of work for a certain cost, and now those costs seem to keep rising, less, you know, it's less work being done and more tokens being used.
• 13:22
  SPEAKER_01: So what do you see in the future as far as is that gonna even out soon?
• 13:28
  SPEAKER_01: Um, you know, what do you see going on there?
• 13:31
  SPEAKER_00: So I think that uh uh I think that in absolute terms, token uh usage across the economy is going to continue to increase dramatically as uh as AI touches more parts of the economy, um, as more and more uh use cases, applications get built out that we rely on uh uh on AI.
• 13:49
  SPEAKER_00: At the same time, I do think that there's going to be a shift from, let's say, a relatively expensive token to less expensive uh tokens as organizations get through a phase of, let's say, experimentation and first deployments and start moving into a phase of optimization.
• 14:06
  SPEAKER_00: You know, basically what we what we've seen today is organizations using the uh the top-tier models to test, to prototype, and to build essentially the first production versions of their use cases.
• 14:18
  SPEAKER_00: Now, what they're doing is going to that second step and saying, okay, well, how can we get equivalent performance or performance which is 99% is good, you know, whatever that threshold is that the that they need for that given use case while optimizing for cost.
• 14:31
  SPEAKER_00: That might be moving to a you know smaller or older model from uh from a same or similar provider.
• 14:37
  SPEAKER_00: In many cases, it it looks like organizations moving to open source models, uh, and in some cases, open source models running on infrastructure that they manage themselves, um, which of course is uh um is the uh uh the architecture, which guarantees that nobody else can cut off your access to that model.
• 14:56
  SPEAKER_00: And so we're in a phase right now where organizations are setting up that capability to one optimize um and then move to smaller, more specialized models and increasingly open source models.
• 15:09
  SPEAKER_01: Well, looking toward the future, if there was one key takeaway you could leave our audience with today, what would that be?
• 15:16
  SPEAKER_00: I think that it's that uh, you know, boards are asking CEOs to defend the AI outcomes faster than companies can actually explain them.
• 15:25
  SPEAKER_00: The operational reality of AI is colliding with the governance expectations around it in uh in real time.
• 15:31
  SPEAKER_00: And we're in this really dynamic phase where we're both building, learning, optimizing, and learning to govern all of this all at once.
• 15:39
  SPEAKER_00: And so I think that this is going to be a period over the next 12, 18 months of a lot of not, I wouldn't call it uncertainty, but a lot of dynamic movement within uh within the marketplace um as we're all figuring this out together.
• 15:54
  SPEAKER_00: Um, so I guess the you know, the the the leaving uh you know the last message is um both buckle up, but also look around you.
• 16:01
  SPEAKER_00: This is an incredible time to be part of this space.
• 16:03
  SPEAKER_01: It definitely is.
• 16:04
  SPEAKER_01: It will be exciting to see the future unfold and it's it's happening so quickly.
• 16:09
  SPEAKER_00: It certainly is.
• 16:10
  SPEAKER_00: Uh, and you know, it's a uh really a unique opportunity, I think, in the uh really the history of uh uh of technology and its adoption but uh in the economy.
• 16:19
  SPEAKER_00: Um so yeah, I think we're uh as as crazy as it as it feels sometimes.
• 16:24
  SPEAKER_00: I think we're all very lucky to uh to be here doing what we're doing right now.
• 16:28
  SPEAKER_01: Most definitely.
• 16:29
  SPEAKER_01: Well, thank you, and thank you to our audience.
• 16:31
  SPEAKER_01: And if you have any questions or comments, please leave them below and I'll try to respond as soon as possible.
• 16:36
  SPEAKER_01: Have a wonderful day.