Competing in the Age of AI with Karim Lakhani

I'm Martin Reeds, chairman of the BCG Henderson Institute and I'm joined today by Professor Karim Lakhani of Harvard Business School, who's co author of an exciting new book called Competing in the Age of AI. So thanks for joining us, Kareem.

Happy to be with you.

So maybe we could kick off by asking you to share the basic thesis of your book. Kareem, you seem to be talking about a new way of competing. What is that new way of competing?

Thank you, Martin. So the basic pieces of the book is that there's a ton of hype around AI and a ton of hype around platforms and networks and so on and so forth. And we wanted to sort of take a step back and say like, what's new and what's different about these AI first companies, these AI powered companies and what's old, what's new and old and so that we can actually create a framework for analysis. And in the end, three things stand out for us. The first is that while there's, you know, when we think about AI, we think about it from the Star Trek perspective, you know, the computer answering all questions, this universal computer working away. And that's sort of in computer science world is known as strong AI. But the reality is that most of the AI force companies have what's called weak AI, which is algorithms that are specifically designed to do one thing and one thing alone. They're very narrow, narrowly task focused. Many of these AI firms have an amalgam of these algorithms put together, but each of them specifically is very targeted. So weak AI can go a long way to help you. And the perspective we say is weak AI is important. And then how you think about your business model changes, how you create and capture value is now changing because of the ability of these algorithms to, to be available. And also your operating model, how you design your learning systems, how you achieve scale and how you achieve scope. Both of those, all three of those things are also changing. And that's the main thesis of the book.

Yes. You have a wonderful chart in the book that shows how these new, I think you call them AI factories, these new companies compete. And you basically talk about three things, scale driven by network effects. So very strong scale effects. You talk about scope, which is this doesn't respect industry boundaries. And you talk about constant learning as a construct. So I'd like to ask you, could you contrast that with the best of traditional companies and highlight what's different?

Absolutely. You know, I think, you know, you know, Al Chandler taught us about scale and scope. You know, Pulitzer Prize with new business books. Very few business books wins win Pulitzer Prizes and has did in the last century. And the, the notion before was that we want companies to focus. We want each sbu, each BU to focus and get their work done effectively for the customers that they have. So in the past we would set up different business units, different divisions, and they would be incentivized organizationally and managerially to go go after and figure things out. Even in a multi uniform or even in a single uniform, you have these focus subunits that are going after things, things. And that enables you to deliver great value to your customers. But over time, scaling them becomes difficult. You have sort of administrative inertia. You have the roadblocks and so forth amongst the organizations from cooperating. And what emerged were these silos of business units and data and IT and infrastructure that really didn't talk to each other at all. In contrast, what you have and that model worked, and that model worked really well and brought us to today. And many of the firms that we hail as effective were built on that structure. What we see different in these AI first companies is that there's a perspective that the customer view is universal. It's no longer that you have five interactions with five different business units and those things are isolated. In fact, we see universally. So the data is aggregated across the board. We are often thinking of ourselves as a platform where we do some manufacturing of products and services, but we have a large ecosystem as well available to do that additional value creation as well in that.

So I'm hearing that some of the big differences are then less siloed within and between companies. Even more emphasis on scale.

Yes.

And not a static advantage, but a continually renewed advantage and emphasis on learning.

Absolutely. How this all comes together is that every single interaction with your customer, with your clients, generates data. And that data is available across the board inside of your organization, so you can get better and better informed about what the customer needs are. And then you can then offer them more and more things across the spectrum of activities that they are engaging with you in your industry or across several industries. That view of increasing scale is the core aspect that we sort of see in these new AI first companies. And then the scale then also feeds on itself because as you get more scale, you generate more value for your users, but you also learn better about them as well. And then you can expand scope along.

So when I read your book, I anticipated we'd be talking about technology. Seems to me that you're more talking about the innovation in the operating system than so much about technology. Is that right?

Yeah, I think for me the technology is a given. Many of these AI tools are actually available on demand. You can get them through your favorite cloud provider. Very sophisticated AI tools are available this way. The common example I use is that recently I published an article in the Journal of the American Medical Association Oncology Branch about AI empowered ration treatment planning. The last time I took biology was in 10th grade and here I am publishing in JAMA Oncology because we can now create through multiple ways the algorithms. The key though is the operating model. Can you set up an organization that can actually exploit these things and rethink the ways in which you deliver value and you capture value? That I think is the biggest challenge. I think the technology is in many cases easily available writ large to lots of companies that want to make those investments.

One question that your book raised for me that I couldn't entirely resolve is there's an implied imperative that traditional companies should embrace this way of competing. And that triggered a couple of questions for me. One of them is is it realistic for incumbents to actually embrace this entirely different way of competing? Secondly, whether they can out compete the leaders on these operating systems, even if they do. And thirdly, and maybe this is wishful thinking, whether there are certain businesses or models where a more traditional structure might actually be advantaged.

Great question, Martin. It's fascinating. So I think traditional incumbent organizations have no choice but to embrace this. Either you're going to be competing against a platform or you're going to be part of a platform that is going to be AI first. So even if you're going to be part of somebody's AI first platform, you also will need capabilities in being AI first and network first to be able to compete. But more generally, let me give you an example for Harvard Business School. So we are a traditional multi product firm. We have MBA program, we have executive education, we have a doctoral program, we have an online program, we have publishing all these, all these silos. None of these silos talk to each other. There's no data. So if you did an MBA with us and then you came back for exec ed, we have no clue. If you read an article in hbr, I can't connect it back to were you ever a student with us or not and should I give you some more information or not and so forth. So there's massive data silos. At the same time, we're finding that being online for us matters being, you know, providing an integrated experience for our alumni and Our students that are participating matters and that we want to actually have larger impact and have scale. So HBS itself is going through this like this metamorphosis of saying how much of it is about what happens in the classroom versus how much of it happens online versus do we even have the data structure and the organizational structure to support this new world? So, and then, you know, is Coursera a competitor? Is LinkedIn a competitor for HBS? Eventually, you know, so we're also trying to figure this out. My sense is that organizations will face a choice which is either I build my own thing and make it work, so what Walmart is trying to do against Amazon, or I'm going to go join one of these platforms and learn how to play with these multiple platforms. And both of those will be, I think, a competitive choices they'll face. Just as you look at in the US the scale of what Amazon is trying to do in terms of finance, in terms of healthcare and so on and so forth, you can see that this is a long term play for them and they will start to enter many of these markets that we thought previously were disconnected. So then the traditional companies will have to figure out a response to that. No matter what can they win? I think it's tbd. I think it's still early days. Many of these platforms were initially built with the perspective that it's winner take all businesses or winner take most businesses. It's not clear that that's the case in every single industrial context. If you look at ride sharing, ride sharing is a very local business, it's not a global business at all. And the benefits you get in being at scale in New York or Boston or London or Jakarta are very different from being universally available. So depending on the context, you might find that excellence in local markets matters a ton more than a global footprint. And that might give you advantages and ways to work on.

So that implies some sort of major transformation. Most companies I now know have a digital transformation. Some of them have an AI transformation. Does it make sense to think about a digital transformation or an AI transformation? It sounds like a holistic transformation of the operating model that's required. And I'm wondering, can we say anything about, from the experience so far about the wrong way to do that and the right way to do that?

So I think both the AI transformation and digital transformation are viewed more as tech solutions than organizational solutions. And I think that's where companies run into trouble because, as I said, relatively easy for us to create the technological solutions and create great algorithms that will Provide customer value. In order for us to take full advantage of that and scale it across the organization, we need organizational transformation and operating model transformation. I think it's a misnomer to call it digital transformation. It really is organizational transformation. And that's where executives need to actually understand that the architecture of their organization will mirror the architecture of their technologies. And you actually have to change the architecture of the organizations to take advantage of the new architecture of the organization. So Back to the CEOs, back to the C suite, back to the executives. You have to view this as much of this as a tech tech work as it is organizational work. And those two things are embedded. And then of course with that is a strategy layer as well. What is the strategy that you're trying to achieve and do you have the organization available to make this work or not?

Let's move on to strategy and then to people. So coming on to strategy and then people, you seem to be implying that this is not just an operating model change, but companies need to compete differently. So therefore they presumably need to think differently about strategy. So whether you're the chief strategist or the CEO, when do you need to reprogram how you think about strategy?

The reprogramming strategy actually is very important. And I would say that the things of competitive advantage, differentiation and position, the market, those elements don't go away. They're still core parts of what you do. But now we need to think about the benefits of how we build in network effects into our products or how do we compete in an era where network effects are driving much more value creation. That's a new era, new thing for us to consider. We also think about how sort of the personalization aspects of your customer experience needs to be done as well. And that's not just in the consumer business, but even in the B2B business. You need to think about personalization as part of the ways in which we go after things. So these dimensions are elements that are now part of the conversation around strategy. Similarly, I think we need much more imagination. I think this is very difficult around value capture. I think for most companies, value capture is I make a product and I sell it to a customer and it generates value for a customer and I capture some part of that value. What we've seen with many of these AI force companies is that there are now multiple ways to capture value. I can capture value based on this price greater than cost curve. I could also make it in terms of outcomes based business model. I can base it on some kind of a license Fee, I could base it on many different ways. And part of the confusion I see happening with executives is that in the economy, many of these both startups, but also adjacent companies are doing recombinant innovation around business models. On value creation and value capture versus we're stuck. We create value by focus and we capture value by charging you a price and getting a premium versus all these other companies are trying many, many different things. And that is going to be like strategic experimentation I think becomes a core order of business for CEOs and chief strategy officers.

So that actually sounds quite profound. So in terms of value creation we have this higher emphasis on personalization and, and change in terms of value capture, more biasing towards ecosystems and indirect monetization. On scale, you know, thinking about network effects, not just supply side economies of scale, on scope, you know, thinking beyond the industry boundary, the delusion that we're in an industry and on learning, you know, constantly renewing advantage, you know, not. So this is, this would be quite profound. At least you had the mantra of sustainable competitive advantage. It sounds like you're more talking about serial temporary advantage.

Absolutely. Because I think the barriers to entry, I think in many of these AI force companies has dropped dramatically and then they are rapidly rethinking the boundary structure anyway. But the barriers to scale have increased tremendously. And so for incumbents that have some scale, that have a balance sheet, that have customers that have some knowledge about a particular industry, the opportunity in front of them to sort of reimagine their businesses is actually great. It's huge. It's a matter of being able to do it. And as you know, that's often like the stumbling block at the board level, the CEO level to say am I going to be the one that's going to pull that trigger? Or what is the right way for me to go after this and figure this out?

So that suggests a new type of competitive advantage, perhaps if participating is trivial, but doing it is very hard. This sort of dynamic learning.

Yes.

So maybe it's not static scale anymore, it's the capability of doing, the capability of coordination.

I completely agree with you. I mean, I think so you look at a company like Rakuten and what Mikitani has done in that company or several, many times, both in terms of the cultural transformation of the company and now. So this AI transformation of the company, that gives you an example of a company that is constantly changing, adjusting, adapting and learning to figure out the environment as it's rapidly changing.

So let's come on to people and I'M not talking about will the AI replace the humans? So my question is, in this new operating model construct, what do the humans do? Because there seems to be a key role for technology. So that sort of begs the question, what do the humans do? What does the tech do? Have they interface. How do you think about the humans plus tech equation in this new operating model?

Yeah, I'm going to bore something I learned from Peter Domingos. Peter Dominguez wrote this great book called the Master Algorithm. Computer scientists out of University of Washington now working on, I think as an advisor to a hedge fund as well. What he said is machines and AI won't replace humans or managers, but managers with machines and AI will replace managers without machines and AI. And so the point being that this in many ways in the foreseeable future is going to be a huge competitive advantage. Those people, those managers, those executives that figure out how to deploy this capability inside the firms will be at a much better advantage than those that can't and that that transfers both personally, but also at a firm level as well. And so I think my sense is that just as when people come to HHBS in the MBA program or even in our comprehensive leadership programs, and they learn accounting and they learn finance, and they're not finance or accounting specialists, nobody comes to HBS to become an accountant, right? They learn, but you need to learn about accounting and finance, understand how business works better. My sense is that understanding of statistics, of programming and systems is now going to be as critical for us because it's not just about what it enables you to do today, but how you can reimagine your operating model and your business model model. That's going to be critical. So that becomes a key driver from my perspective as to what needs to be done.

So I can imagine there are many executives out there that are thinking, I understand this sounds huge. I have no idea where to start. So what shall we do on Monday morning Conversation? If we wanted to make a start on this agenda, what's the first conversation we should be having with our management committees?

So I think as an academic, of course you need to learn more and understand more. I would say there have been lots of pilots that have been done right, and most pilots don't scale. So understanding how do you scale these pilots is a core, core thing and keeping your attention on it, right? Saying, I'm actually going to keep attention on these pilots and I actually have a scaling plan. So don't just do the pilot, but bring with me that if the pilot succeeds, how you will Scale, I think is a critical element.

So not just study, not just study and try, but study, try and scale underlying.

Yeah. And you know, in fact you want to actually say, give me your implementation plan. Let's imagine this thing works. We get great benefit from this algorithm or that algorithm or that initiative. Okay, how will we scale and what does the scaling mean for our organization? And I think forcing that issue from the top down is going to be critical because it is going to be in the. The scaling will require org transformation. And I think that was where the rubber hits the road and that's where the top level group has to intervene and make that work happen. So I think that's like so lots of pilots happening. Don't waste your time just doing that. Think of scaling and driving that. The second thing is to actually step back and reimagine and say, how might the current AI first giants enter my company, enter my space and am I being too myopic about their play in our industry? And how might I respond and how do I get ready for that now instead of waiting for that to occur? Because in many ways the benefits of scale and scope and learning that they have from their data platforms compels them to keep expanding. It's inevitable in many ways, unless of course there's regulatory pressures and so on and so forth that come through. But the sheer dynamics of the business world would drive them to that. So do that war gaming, figure it out and then start to think about the transformation you need to do inside of a good company to be ready for that and to be ahead of it instead of waiting for them to show up at your doorstep.

So given your response, I can't help but ask one more last question, which is I see some sort of titanic battle between the establishment AI factories, the new AI factories and the incumbent economy. And to a large extent there are things incumbents just have to do, as difficult as they are. But as they enter that battle, what sort of advantages can they draw on so that they're not just pay limitations of the things that they're trying to imitate.

That's a really good question. I mean, I think the Nokia versus Apple and Google Android is a very instructive story because people often say Nokia's consumer business got killed by Apple because they didn't invest enough, they didn't have the technology portfolio, they didn't have the patents, and so on and so forth. The reality is that most of these companies, Nokia had invented all the core technologies before Apple ever did, had all the core patents, had an App Store before Apple had an App Store. Yet they weren't able to change their organization to, say, single platform. Right. User experience matters. And then it's all about network effects and learning effects from building the App Store. Incumbents have technology, have balance sheet, have people, have customers, have brand. And the question becomes, can you take those components and reorganize yourself to be ready for this world? I think that's the most pressing questions that incumbents face, because I think the smarts are there. People often detect these changes beforehand, but willingness to move at the top level is often missing. I think that's something that we actually have to figure out.

Well, this has been a great conversation. Thank you, Karim, for sharing your ideas with us.

Thank you, Martin, very much. It was a lot of fun.