What the AI Acceleration Curve Means for Corporate Strategy

Steve Odland: Welcome to C-Suite Perspectives, a signature series by The Conference Board. I'm Steve Odland from The Conference Board and the host of this podcast series. And in today's conversation we're going to talk about where we are with AI, more importantly, where we're headed.

Joining me today is John Metselaar, the Council Director for our Innovation Leadership Councils throughout Asia and Europe. John, welcome to the program.

John Metselaar: Thank you so much, Steve. Pleasure to be with you and I'm honored to be invited.

Steve Odland: Yeah. John, everybody knows what AI is. They're two letters. They're A, and they're I, right? But they really stand for artificial intelligence. Lots of stuff has come out in the last couple years. Things are changing but how do you define artificial intelligence?

John Metselaar: I like the fact that we started fundamentals, actually. When I teach I get back to definition before we jump into things that we don't fully understand. The essence of AI really is about representing encoding pattern recognition at scale. And as a consequence, it allows to process vast data sets that identify correlations that we might never perceive. This contrasts with human intelligence that excels at more nuance judgment within specific boundaries and contexts.

Steve Odland: At its core, John, it's software, right? It's still ones and zeros. It's the next phase of the digital evolution. And that's something you've been talking about for nearly 40 years, the digital evolution. This is just the next frontier, isn't it?

John Metselaar: Yes and no. What sets AI apart today is the ability to learn from experience rather than follow predetermined rules. We're observing a shift here that software does not only follow instructions but can start to infer things and look at constant improvement, constant learning and, therefore, unleashing huge potential.

Steve Odland: You're right. You hear all these terms, John, with AI. Anytime you're launching into a new frontier people use different terms. But in the business world today, you hear mostly about generative AI, agentic AI, machine learning, large language models, robotics, and so forth. Can you just talk through these forms of AI being used by business today. Let's just start with generative and agentic.

John Metselaar: Let's start indeed at generative AI. It's mostly, at this point, about generating things. Today it is mostly used as a sophisticated drafting tool. Marketing teams generate content variations. Legal departments use it for contract templates, software developers get help in coproduction. But it only scratches the surface of its potential.

The more transformative developments that involve using generative AI, you have to start thinking about rapidly prototyping solutions across domains, number one. Number two, testing product concepts through synthetic customer interaction. Don't run through very long real customer experience anymore but get real quick data and results through synthetic interactions.

And then maybe the biggest one is exploring strategic scenarios by generating different market simulations. There's also things like developing training materials that are tailored to individual learning patterns, for instance. The key here for implementation and for success is to connect that strength of AI with that human refinement, rather than going for the finished product.

Steve Odland: So generative AI is using the tools to generate something. Content, you said, whether it's written or audio or video or the creation of something. And the next term that's used a lot is ?agentic AI, which has the word agent. So AI is acting as an agent or as an automated tool.

John Metselaar: Absolutely. The key word here is autonomy. This is the frontier that we are now. This is very hot in today's environment because we're really starting to touch on that frontier and play around with it. And this is where systems operate with increasing autonomy towards very clearly defined objectives rather than executing specific instructions.

So, for instance, right now early developments on this, just to bring this to life a bit, procurement agents. They're able to negotiate contracts with predefined parameters-- that remains an absolute critical limitation. Research agents that synthesize information across databases and then they present those synthesized findings to the humans to be integrated with that judgment that I mentioned earlier.

And then things like operating agents. They monitor systems and implement corrective actions. They can monitor an incredible number of parameters that humans can never do. But again, within established guardrails. So the critical discipline for business leaders here becomes to define the scope of the autonomous action precisely, clear boundaries, and then build robust monitoring in there.

So this is a relatively new frontier, as we both mentioned. And I would say, try and move forward with caution in general. Get experience with the new AI systems. Learn then constantly refine. It's absolutely critical that you built that experience within an organization because, I think I heard you say that as well at CNBC, others are doing it as well. There is this competitive pressure that is definitely at play.

Steve Odland: People don't want to get left behind. And when you're starting off into a brave new world, you don't know what it looks like, so you have to rush ahead and try it. A couple other terms, as I said, that are being bandied about are machine learning and large language models. Where are they?

John Metselaar: They are the more basic elements of AI. We've gone beyond that with generative AI and with ?agents. Machine learning is you learn as a machine and that can be turning into language.

And, by the way, it's interesting. We are almost three years from the launch of ChatGPT. And, of course, that was the first time where the connection with machines becomes intuitive and not through any difficult coding anymore. So large language models really have sparked that hype that we're seeing today. And development is moving really quickly.

Steve Odland: It's interesting how much ChatGPT is being used as a more sophisticated search engine. Google was very sophisticated at one point. It gives you a list of websites but then you have to plow through all these websites.

Using ChatGPT, it comes up with an explanation, the answer that you're looking for. May not be precise, it's got issues, sometimes it hallucinates, meaning it's got wrong stuff in there. But that's not different than any website. You have to check your sources and so forth.

But it is interesting how these tools, as they're introduced over time, get used in a way and they settle in. We've seen that in the digital evolution over the decades.

John Metselaar: Yes, we have seen that in the digital evolution. You mentioned that, of course, we've seen this in the nineties, right? And so there is some similarity in terms of moving things forward. This is the next wave, you can argue. Having said that, though, I do believe that AI's economic potential can exceed that of the digital transformation while facing different constraints.

Because think about it. Digital transformation primarily eliminated friction in specific areas. Communication delays, information silos, breaking through those manual processes that got automated. Things got easier. AI promises something more profound. It promises enhanced decision quality across an entire organization.

That's the potential but that's also the challenge for organizations. Leaders need to start to think very boldly about where they want to take their organization, their company, fundamentally re-imagining operations when human processing capacity now no longer is limiting organizational courage and structural flexibility.

It's not so much the technology that is holding us back right now; that's advancing quite nicely. We're advancing really quickly. But it is more what I call the institutional inertia and the political complexity of genuine operational transformation.

Almost like you want to get to new business models but your operating system that you developed over so many years is holding you back. Chand, one of our senior fellows from Cambridge, is calling this business model incoherence, where the business model and the operating system are now out of whack.

How do you go and reinvent yourself? Absolutely critical years breaking through these silos. That is the promise and the challenge to fully tap into the potential of AI.

Steve Odland: Yeah, it's all true. If you go back to thinking about AI as a software, all the things that you just described were using software tools. But it also enables new technologies on the hardware side. As an example, robotics. In manufacturing facilities, AI is being used as the backbone and the brain, if you will, behind the whole robotics introduction. So there's a complimentary hardware, software evolution going on as well.

John Metselaar: I think that's another part of the potential of AI, indeed, that you connected with hardware and leverage robotics in so many different places for so many different purposes. Of course, everybody has seen the Amazon robots going around in that very complex matrix without colliding. That's so impressive. But again, we're still at the cusp really of the full potential here.

Steve Odland: Where are we on the development cycle with AI? Is this just stage one? And then, how many stages are there?

John Metselaar: We need to recognize and we need to deal with and we need to tap into, frankly, the fact that we're conditioned to think linearly. But that can be very dangerous when you look at new technologies because they typically follow, I should say, an exponential dynamic. The patterns show really slow initial progress followed by surprisingly rapidly mainstream integration. We've seen that in the past, like the point that you made earlier around digital transformation.

With AI, I believe this dynamic can be particularly dramatic because of that capability for self-learning. Each successful implementation creates new learning that accelerates subsequent deployments. When you go for technical challenge in the past, you would fix it and then move on to the next technical challenge. Now, when it gets solved, you can apply it at scale across different use cases.

The organizational resistance that is still in place, sometimes at all levels of the organization, depending on the company and the individuals, that will diminish, right? Familiarity builds, value gets demonstrated.

Think about all the stakeholders that you deal with, your suppliers, your customer base. They will start to use the system as well. So integration with them is going to be facilitated and smoothened and, therefore, allow a much faster progress. These are real factors.

And then again, we mentioned already the competitive dynamics. Once peers demonstrate significant AI-driven advantages, lagging organizations will face existential pressure. I always say that organizations that plan AI strategies based on conservative slash linear timelines, they are preparing for yesterday's space rather than tomorrow's reality.

Steve Odland: Yeah, and it's interesting. In The Conference Board's latest CEO confidence survey, we asked CEOs their views on AI and they told us that their biggest worry is being left behind, as we mentioned before. Their primary objective with AI investment is to make sure that doesn't happen.

And then when you ask them how they intend to deploy it, it's mostly around cost reduction and operational efficiency. Therefore you hear this thing about, oh, all these jobs are going to get taken away by AI. What's your view on that? If you think back to the eighties and nineties and the deployment of the personal computer and Microsoft Office and all the tools that came with it, yeah, there were some jobs that were destroyed, but a lot more created. And a lot of people got more efficient.

John Metselaar: I don't know, Steve. My speculation is that I don't think we're going to see so many jobs being replaced by AI. As so many people say, it's not AI that's going to take your job, it's the person who is able to deal with AI that's going to take your job. And I think that's a better way to look at it.

I think AI will have a huge impact, as I said earlier, I think already about the middle management. How does middle management, the task that it does. Many of the tasks can be delegated now already and in the future, even more time-consuming, tedious tests can be delegated. Middle management, for instance, can be freed up much more towards higher-value opportunities. As a consequence, they'll play a key role in identifying solution spaces that used to be way too vast for traditional methods.

And this shifts advantage. This is interesting. I think this shifts the competitive advantage towards organizations that can formulate the right questions and evaluate AI-generated possibilities rather than those that keep doing what they've been doing over the last years.

I was actually just listening to a YouTube video yesterday. One individual that I follow was saying, he was very wise. He said, become unbelievably curious at all levels of the organization, unlearn, become unbelievably detailed, learn how to prompt, give the right guidance, and become unbelievably human. Because that's where the compliment with AI lies, right? And the interface is now so natural. So he was concluding those three points by saying, make AI your thinking partner. And I think that will be hopefully, not only serving businesses, but also humanity.

Steve Odland: We're talking about the state of AI in the world. We're going to take a short break and be right back.

Welcome back to C-Suite Perspectives. I'm your host Steve Odland from The Conference Board, and I'm joined today by John Metselaar, Council Director for the Innovation Leadership Councils at The Conference Board in Asia and Europe. So John, we were talking about what CEOs are saying about AI. Another objective for CEOs is to help them drive innovation in new products and services. Give us some examples of where you see this happening.

John Metselaar: AI is indeed being used in developing new products and services. That can range from AI-powered products in their own right. Also, maybe even more importantly, at least more broadly, I believe, is AI can help and maybe even redefine how we innovate, as we used to call it the P and G many decades ago.

Just a few illustrations for this. Pharma, they employ AI to go and identify promising molecular structures on a timeline that is now days versus years of testing in the lab. Financial services firms develop hyperpersonalized offerings. They synthesize individual behavior patterns with market dynamics at scale. Manufacturing organizations optimize product designs because they take those millions of parameters and they create combinations in ways that human engineers in the past could never practice and could never evaluate. These are a few examples.

It really is about AI enabling exploration of solution spaces too vast for traditional methods. That's the way to think about this. It comes back to the point I made earlier, which is it's all about learning. And so curiosity, become detailed, be human and then create this learning organization. That has to be the norm for the future.

Steve Odland: Some people are worried about AI sentience, meaning AI actually becomes more human and takes over and basically overtakes humanity, even. It sounds like science fiction but there are groups that are urging governments to put restraints and controls on all of this. What's your view on ?sentience? How close are we to that being reality and what are the risks?

John Metselaar: This is a hard one. I think we all want a crystal ball on that one, to be honest. What is clear today is that there's no meaningful ?sentience by any reasonable definition. For now, AI is, as you mentioned, software, a very sophisticated software, very capable software but software. And it process patterns without understanding and generate responses without comprehension. That's what it does to you and that's where you need the human judgment to come in and that's where the power lies if you make that connection proper.

I'm a bit worried that whole discussion around ?sentience takes attention away from more immediate governance challenges. I see the relevant concerns here center on three things: concentrated power, who benefits from AI; systemic bias; and the erosion of human judgment, as it overrelies on algorithm and what the software spits out. These risks, they're very present already now and they will only get worse and, in my view, they require institutional responses.

So yes, I think you do need to go and develop regulatory frameworks for high-stakes applications. I think you need absolute transparency for consequential decisions that affect mankind, in particular. And then you need to preserve human agency. I hope we're not getting lazier through AI. I hope we use it in the way that I described earlier, as a plus, to augment our capability. And I think that will help humanity be better than rather than worse.

Steve Odland: I don't hear any fear in your commentary. Is that mainly because we're so far away from it or because you have faith that these systems will be guided by proper judgment and rules as we go?

John Metselaar: As I said, the excitement here is about genuine enhancement of human capability. It will extend our analytical reach, accelerate scientific discovery, hopefully for the better of humanity. Also, again, that use complexity that AI is now able to grapple with and deal with and generate insights from.

The potential is there and I think it can be very powerful. The worry I see is honestly is what I said earlier. It's not so much the technology itself but it's more how we are able to work it ourselves. As leaders-- political leaders, business leaders, policy leaders-- how we organize around it. Will we be able to have AI augment human judgment or replicate it? Will we able to distribute these benefits broadly? Or will they be concentrated narrowly and a few gain and a few start to rule what we all, what everybody else needs to do?

And then the governance framework. We do need to establish governance frameworks that ensure accountability, transparency, and fairness, while still allowing innovation to happen.

These questions, in my view, they demand immediate attention from institutional leaders . That's urgent because technology evolves rapidly and our capacity for thoughtful governance doesn't always follow.

Steve Odland: These are really important points. Let's go back to the company because, most of our listeners are in companies or in leadership roles. As we've polled executives, we find that AI has a head start in human resources (HR) functions and marketing functions. HR has been using machine learning, large language models-- the fundamentals, as you said, the building blocks-- for a long time. Marketing is now using generative AI more. Give some examples of these areas and then where are the next deployments? Where's the most fertile ground?

John Metselaar: HR and marketing, these functions, they're particularly amenable to AI precisely because they involve processing large volumes of semi-structured information toward probabilistic outcomes.

HR applications, take resume screening. They identified qualified candidates much more consistently with much less manpower than a manual review or than multiple manual reviews, obviously. And they can analyze employee sentiment for early signals of retention risk that can tap into different communication channels and sources.

Take marketing. They deploy customer segmentation at granular scale. They can optimize campaigns across channels in near real time, huge advantage. Or content personalization, that we're seeing already happening now in our daily lives, has become a bit of a norm. But you can adapt the messaging to individual engagement patterns.

These are obvious areas. We have to be really careful that we, again, constructively bring together those data sources with human judgment. I've seen several instances where that led to mistakes, so don't delegate decisions to the machine. That is true for HR, marketing, but even beyond as well. Again, the strength, the potential lies in connecting the machine with human judgment.

Steve Odland: Really words of wisdom there. Some people have said that whatever you think the rate of AI is going to be implemented, really just double, triple it. And this goes to your point about, it's not linear is going to be exponential. Why is that? Is it just that the technology's going to get better or is it just that as we get into this thing, it's the natural learning curve and people just start adopting at a higher rate?

John Metselaar: I already touched on that. It happens with every technology, right? When it comes in, they follow the S-curve, as they call it, in the innovation practice. So slow, initially slow, and then you start to get to the exponential part of the S-curve and suddenly, things happen really quickly until finally, after many years, technology wears out and the S-curve flattens. Where we have been over the last several years is still on the relatively shallow side of the S-curve.

Through experience, technological advancement, but also as I said earlier, through organizations learning how better to use it and make some of these bold organizational changes. Modify your operating system to be more consistent with the new business model that you want to get into. With AI this can be particularly steep.

And that's why again, folks need to step into it urgently if in business because of the self-learning capability of AI. That allows that acceleration to go faster, in my view, than what we've seen in the past.

Steve Odland: Last question and just to wrap up. When you look at the potential of AI, and you've been looking at digital transformation for a long time, I do view this as the next phase. Just it's more digital transformation in a sense. It's different, of course, each phase is. But one of the things that excites you and what worries you?

John Metselaar: What excites me. It offers the potential of better decision-making through more comprehensive information synthesis. As I said earlier, analyzing the incredible complexity of data and then combine that with human judgment. That's one. There's better decisions, particularly if you're able to go beyond the operational application in specific departments, are able to go and really deploy and employ and leverage AI at an organization level by breaking through the silos, and are able to reinvent your organization.

Second is faster innovation because you can much more quickly explore solution spaces and then, hopefully, more equitable outcomes. Because we inevitably have biases. And if we get to transparent AI systems and that is in my view, a must, then we can actually get to better outcomes and less bias. But that obviously requires the input to be unbiased and transparent.

Steve Odland: Alright, we're going to leave it there. John Metselaar, thanks for being with us today.

John Metselaar: It was real fun. I hope you found this helpful. Thank you.

Steve Odland: And thanks to all of you for listening to C-Suite Perspectives. I'm Steve Odland and this series has been brought to you by The Conference Board.