MOD: Masters of Data

Bringing the human to the data

Marco Iansiti and Karim Lakhani: Competing in the Age of AI

Marco Iansiti and Karim R. Lakhani, Professors, Authors

2020年03月09日

45:32

[With AI] I can serve a million people just as well as I can serve five [...] I can learn from the million people that I'm serving.

Marco and Karim talk about their groundbreaking book on how to use AI to supercharge your business, compete, and win

Show Notes

Ben Newton:

Welcome to the Masters of Data Podcast, the podcast that brings the human to data. And I'm your host, Ben Newton. One of the most controversial and discussed topics today is artificial intelligence or AI, and the implications of it. Is AI good or bad? Are we being replaced? Is AI exacerbating discrimination? Our guests today are taking a much more practical view, one that seeks to capture the value of AI for business and help corporations large and small take advantage and compete.

Ben Newton:

Our guests today are Marco Iansiti, the David Sarnoff Professor of Business Administration at Harvard Business School, and Karim Lakhani, the Charles E. Wilson professor of Business Administration and the Dorothy and Michael Hintze Fellow at Harvard Business School. They have just released the book, Competing in Age of AI, which we're going to talk about today. So without any further ado, let's dig in.

Ben Newton:

Welcome everybody to another episode of the Masters of Data Podcast, and I'm very excited about our episode today. As always, we try to bring interesting people on the podcast, and I think we've succeeded beyond our wildest dreams today with our two guests. So we have two professors from the Harvard School of Business. We have Marco Iansiti and Karim Lakhani. Welcome onto the podcast. It's good to have you here.

Karim Lakhani:

Thanks, Ben. [crosstalk 00:01:30]

Marco Iansiti:

Great to be here, yeah.

Ben Newton:

You guys have a long history in this space, particularly around the things we're going to be talking about today. And I think maybe a good place to start is based on where you guys are coming from, and your experience in the space, what made you decide to write this book? So let's actually say the book here, the book you guys just released, Competing in the Age of AI. I read it. Loved it. I think it's an amazing book for the time, but if we take a step back. Why did you guys decide this is the right time to write a book like this?

Marco Iansiti:

Well, it's an interesting question. We've been at this for a while. We've been, I think, for the last 10 years really, we've been looking at digital transformation or the changing nature of our economy and technological change. Karim and I have been kind of working on this for a while together. We taught classes together. We got in trouble with companies together. Yeah, we love to get in trouble.

Marco Iansiti:

And I think after a bunch of years of doing this stuff, it started to really dawn on us, that more than just individual things, individual events, individual things being different. New use cases, new cool algorithm applications, new cool software. Something really fundamental was beginning to change in the way that our economy works. And the way that firms work within the economy, which, for us at a business school, it's kind of a big deal, because we worry a lot about firms and managers and those kinds of things.

Marco Iansiti:

And then, all of a sudden, it almost felt like there was so much technology out there right now, that there is really critical mass of applications built on it. That there's a whole range of companies where the technology itself, the algorithms, the AI, the digital networks, that make up the technology components were becoming the core of the firm. In other words, for hundreds of years, this industrial revolutions firms have been about people, managers, workers, and then eventually somebody started sprinkling a little bit of information technology to make the managers and workers more productive. And all of a sudden, the firms were becoming almost technology first. If you look at who was doing things, who was making decisions, who was doing this stuff in companies-

Karim Lakhani:

Usually the work of the firm itself. Yeah.

Marco Iansiti:

Yeah, who was really delivering the value of the firms delivered to customers. Increasingly it was the technology and was not the people.

Karim Lakhani:

And a little bit of history on top of this, both of us have studied the software industry and what were the other... Marco had looked at ecosystems in the software industry, looking at Microsoft in the 90s and the 2000s. I became an academic to understand the open source software movement. And the software industry has been through so many iterations. What's interesting to us, was that the iterations we saw the software industry go through were now also happening in other adjacent industries or even far away industries.

Karim Lakhani:

I'd laugh when people say, "Well, Uber's a technology company." Well, it's a transportation company, but Uber's has brought technology to the transportation industry, has made it effectively a software business, right? Same with Airbnb and so on and so forth, or AM Financial in banking. The dynamics of what we saw happen in the software business are now happening across all industries. And then that's a big deal, because the way you structure yourself is very different in a traditional steel company versus in a software company, and now these AI companies as well.

Ben Newton:

Yeah, that makes a lot of sense. Because you have the quote in the beginning of the book from Satya Nadella at Microsoft about AI is the new run time. And I mean that comes up again and again in the book. And do you feel like thing is kind of the next stage of this whole thing that Marc Andreessen said about software eating the world? Because I like the way you put it, Karim, the whole idea that it's not just about all these companies becoming software companies. They're applying the technology and AI and things to a business sector to revolutionize it. So is it kind of the continuation of what Marc pointed out in the early 2000s, or?

Karim Lakhani:

100%. The Andreessen quote is so [inaudible 00:05:58]. In fact, we started this course at Harvard Business School on digital innovation and transformation in 2013, and the first class starts with them reading Software's Eating the World at that time. [inaudible 00:06:13] impression in sort of observing how this was the big driver. And what we say is that software is platforms, it's data, it's analytics as eating up the world. And not just eating it up, but, in fact, reconfiguring the world, reconfiguring organizations in the process.

Marco Iansiti:

Yeah, the basic unit of analysis for us is the firm in this. And I think what's happened, that beyond software eating away at bits and pieces of the firm, and beginning to go from a core of software in the software industry to industries like banking when there's a lot IT across a whole variety of different functions. Or to the hotel industry, where there's IT running a bunch of different processes. The whole firm is turning inside out, where essentially the software is managing the firm in many different ways and telling the people what to do, where the value delivered, fundamentally is software first, or it's AI first.

Marco Iansiti:

It's what Satya means when he says the AI is the run time. AI is the execution engine. We spend so much time worrying about business execution and how you do it better. I think that the idea, is that if you have more and more use cases that are being enabled by AI as the core of the firm, that becomes the execution arm of the firm. Take an Amazon. An Amazon warehouse is a really interesting thing. I mean there's a bunch of people in the Amazon warehouse running around, but the software is telling the people where to go, not the there other way around. So I think it's a fundamental shift generally how companies are thought about, with lots of implications across the board.

Ben Newton:

Yeah. Well, one thing that might be interesting here, again, as I was reading through the book, you guys talked about the challenges. Because I've even done this in my own roles in the past, is trying to set the stage about why companies are applying technology, right? They don't supply it for the fun of the technologies, though, a lot of engineers would want to do that, right? They're applying it because they're solving a business problem, and because they think it's going to give them a competitive edge. And you guys talk about these three challenges: scale, scope, and learning. So maybe talk a little bit more about that. I mean why, going one level deeper, why is AI making such a big difference?

Marco Iansiti:

Yeah, no, it's interesting. That's really what dawned on us as we were pulling together the material for this book, that something fundamental was shifting, not just in the fact that the AI was solving individual problems and driving individual use cases. But when you have a whole bunch of AI there, the whole nature of the firm changes. And so traditional constraints on managing a firm had been removed. Traditionally, since the Industrial Revolution, we tried really hard to manage firms to drive scale, like in a manufacturing plant and you have standardization of work, and people do the same kinds of things over and over and over again. To try to do as much stuff as possible. Or in a bank and processing loans or something like that. The thing that dawned on us, is that when you have software doing these things, when you have algorithms that essentially are defining whether or not somebody will get a loan on AM Financial, then the whole nature of the scalability of the process changes dramatically, right? Then there's essentially zero margin on cost.

Marco Iansiti:

I can serve a million people just as well as I can serve five. In fact, I can serve a million people much better than I can serve five, because I can learn from the million people that I'm serving and gather data, making the processes better. And so, in many ways, it sort of changes the whole concept of scale for growing a firm, which is a huge deal. Scope is the same way. Not only when you have a software-based firm core, as you would see at Airbnb or at Facebook or something like that. You can plug a bunch of different things into that software core much more easily, and so all of a sudden go across industries very, very, easily. And Airbnb can have an experiences marketplace, that plugs into its own system for allocating rooms to people very easily, and use the data they gathered from one application to drive value in the other way.

Marco Iansiti:

And so scale, scope, and then finally learning, it's almost like an obvious one. In the sense that the more people you serve, the more data you get, the more opportunities you have for learning, and the easier it is to aggregate all the information. Much more so when you have a bunch of algorithms and databases gathering all this stuff, then if you have a bunch of scattered people trying to figure this thing out under some sort of a management structure. And so scale constraints, scope constraints, learning constraints are removed from the firm, and you can have these organizations do some pretty remarkable things.

Karim Lakhani:

And by the way, these three dimensions, scale, scope, and learning, are what define the operating model of a firm, right? It's important [inaudible 00:11:04] think about it. That's what companies are actually geared to do fundamentally. And this lesson we learned from Ale Chandler in the last century, when he was looking at American capitalism, and trying to understand why American capitalism did so well, it's because of this focus on scale and scope, that the companies had.

Ben Newton:

Yeah, no and I think the way you guys brought that together in the book, I think I've heard it described a lot of different ways, and I think it really makes sense. And then in the examples you guys use about how companies are waving it in, because we were talking before we have a previous podcast with Stitch Fix, and again, I think particularly listening to their model and some of the other ones I've heard. Of how it's not just about blindly applying the artificial intelligence, and putting some data scientists in a room and say go do something. It's really about using it to increase the scope and the scale of the company, and it's not just about pulling humans. I think you guys, if I remember right, you said something in the book about not just replacing humans, but maybe moving humans to the edges, did I get that right?

Marco Iansiti:

Exactly, no, I mean there's humans everywhere, right? So there are lots of managers and workers at Amazon, there are lots of managers and workers at Google, at Microsoft, at Stitch Fix. And so forth, but I think the way to do it, is in some ways you move the humans off the critical path. In other words, I can serve a customer. I'm a customer of Stich Fix, they know a lot about me. The AI is pretty good, it will generate some options, right? So it will generate a way to get my next box, here's the different things that Marco can get. And then humans curate that, but in some ways the processes enabled is accelerated by the fact that you have this AI, especially severing out the options.

Marco Iansiti:

Then the humans can kind of tweak that a little bit, and make it better. And so the humans in some ways is supervising the AI, in much the same way as it can happen in any other business. Like if I'm on my Amazon account or whatever it is, any expectations can be handled by humans, but the basic operating model can fire off options and predictions, and essentially get things going for the firm. And so that remove the bottle neck. So I can still add value to it as a human being, but I'm not in there in the actually critical path of execution. So in some ways I'm out of the run time. I can still go out there and design it and tweak it and evolve it and check it, and make sure that it's doing things right, but I'm not constraining it as a bottle neck.

Ben Newton:

The way you describe it there Marco, I'm wondering if the... And I know that people have stated this much more eloquently than I'm about to, but basically when we adopt technology in business, it does seem like there's a transition period, where figuring out how to integrate it with the humanity in the companies. To integrate it with the human process, it takes time. I mean like how long it took us to use spreadsheets and word processors and personal computers, because it's not like AI is new. It's just that it's really coming into it's own.



Karim Lakhani:

Yeah, but also Ben I think the thing we note, that if you put AI over existing processes, then you'll just make the existing processes perhaps more [inaudible 00:14:30]. I think key is to redesign the processes, right? To take advantage of the technology. And part of what we've sort of seen in the history of technology adoption in companies, is that those organizations that can transform their processes and their architecture, their organization architecture, to make what the technology enables, they do way better and survive a lot longer, than those that say, no we're not going to adjust. And that's where I think, there's all this talk about digital transformation, and we're just going to go cloud, we're going to add AI, we'll sprinkle some AI magic on our organization. But if you don't do the hard work of actually transforming your organization with it, it's going to be for not.

Marco Iansiti:

Yeah, and I think that's really important. I mean, we make a point in the book that it's really the different architecture for the firm, right? That the traditional firm is a bunch of silos, it's a bunch of individual teams, individual functions, individual business units. That are designed to do things as autonomously as possible, because that's what humans like. You don't want to work with a team of 1,000 people, you want to work in a team of 5 people, maybe a team of 10 people. But as it gets bigger and much more complex and hard to manage, the data doesn't work the same way. Data knows no boundaries, and so the data that Amazon might have on Karim is useful across every function, or every sort of bit of AI, that you can use. And so, it makes sense to create an architecture for the firm that's much more horizontal, and much more of a platform structure that shares everything across business units. And that's the hard part in the transformation process.

Marco Iansiti:

And if you look at firms like at Fidelity or Walmart, that have been working at this for quite some time, their transformation challenge is not so much just deploying the technology, because technology is available, it's on the cloud it's great, it's fantastic. But a lot of it is changing the way the organization functions, and the way that's it structured, and the way that it's culture works, and the value systems and all these different kinds of things, that essentially turn inside out. All of a sudden the technology is running the operations instead of the operating managers.

Ben Newton:

Yeah, I really did definitely, considering my background and having worked in kind of the enterprise IT operations space in particular, the way you guys described that made a lot of sense. Because one of the things I've seen over time, is there's always been a tendency for the last couple decades as oh, we're going to go apply this whatever the new shiny thing is, in technology. We're going to apply that, and things are suddenly going to change, and there's always been an underlying culture bit. It's like, guys you can apply the technology all you want, but the technology was built... Take cloud for example. A lot of the benefits have gone to cloud computing, I think you guys kind of touched on this in the book, is that but it's based on certain changes you make in the culture of your company. And if you just lift and shift and move things there, you're not getting the benefit. And I don't think a lot of companies realize that.

Marco Iansiti:

No and I think the fault lies both with the technologist and also with the managers. So pox on both their houses. I think technologists think, well a rational things, well I've got this new shiny object, of course it's been the universal solvent for everything, and it's going to magically solve all of our problems, so let's go in this and this and this. And managers come back and say, I gave you all this money, nothing has happened, hey what's going on? Because they haven't come together and said, this is as much as it's a technological change it's an organizational change, and we need to do that together.

Karim Lakhani:

It's the reason why so many pilots kind of get stuck. You do a pilot of a cool new shiny technology, right? And the pilot looks great, wow we got all these different results and exactly what we expected, and then everything stops, right?

Marco Iansiti:

Maybe you should do a podcast on all the pilots that have stalled amongst [crosstalk 00:18:32].

Karim Lakhani:

Yeah and it is likely stopped, because all of a sudden people in the organization went, oh my God, this is what we actually-

Marco Iansiti:

I have to change how I work?

Karim Lakhani:

Yeah, I have to change all this stuff to actually make this thing really deploy across the business and work. And all of a sudden it's like, I don't want to do all that. So the polit was great, nice shiny pilot, but let's keep it away from the actual core of the business.

Karim Lakhani:

Right, right. We made the CEO happy, we did what he read in a magazine somewhere, let's move on.

Marco Iansiti:

Yeah, [inaudible 00:19:07] we read all those magazine articles, because we were flirting with it.

Karim Lakhani:

Yeah, exactly. [crosstalk 00:19:09].



Ben Newton:

Yeah, you guys are exactly right. I mean it just came to mind, some of the things recently, with big data and stuff like that. And all the hype and then suddenly it just disappeared off the map. And one thing that was really interesting, that I think you guys did really well in the book, is tying the business models and this idea of how firms work. Like you said the firm's operating model with the technology basis. And you talked about this concept of an AI factory, which I just found fascinating. So maybe talk a little bit about that, because you basically said that's kind of the core of this new technological model. So talk about what that means.

Karim Lakhani:

Yeah, absolutely. So look, again the thesis is that the core of the firm will have an AI factory, and that core impacts both the operating model and the business model. The business model will be defined as value creation and value capture. All the ways in which people come to you and transact with you and the reasons for that is value creation. And all the ways in which you make a profit, make money from that transaction is value capture. So that's what defines a business model. Now what we see is that AI factory helps you both create value and capture value, but also it drives your operating model. Helps you achieve scale scope and learning. And for us, the notion is that of course the input of the AI factory are data, right? Data coming in of all types, all different variety at an unprecedented scale.

Karim Lakhani:

And the data has to be processed and cleaned up. By the way, massive underinvestment in most companies on the data pipelines, right? We just don't invest enough, because all the dirty janitorial work that needs to be done to get the data sets cleaned and ready and integrated, no CIO got famous for creating the data pipeline, right? But that's the problem, right, because there's no attention given to that aspect. So dad pipelines are super important. That fits into an algorithmic engine, that is doing three things. Either making a prediction about the future state of some action that needs to be taken. Classifying events that are happening inside the firm, or automating processes all the way. And all those things are happening as an outcome of the AI factory, and there's an infrastructure system that can take those predictions classification automation, and deploy them at scale. That's the underlying AI factory that we see.

Karim Lakhani:

Now here's the thing that struck us, and was a big aha for both Marco and I. Which was McDonald has a hamburger factory, right? And they make hamburgers at scale, so [inaudible 00:21:48] or big macs. I need to stop going there, but the hamburger factory at McDonald's looks very different from the steel factory that GM uses or Ford uses to make their steel for their cars, very different. But the AI factory at McDonald's is going to look exactly the same as the AI factory at Ford and GM. And that's the key thing, and that factory also looks the same in Google and Facebook and Microsoft and AM Financial and Alibaba and Flip card and so on and so on. And so this notion is that the AI factory is the generic capability that we all need across our companies, to both build our companies, but also to integrate into these various networks that are out there, that are doing this things.

Marco Iansiti:

And that part, I mean it's a huge revolution in management thinking, right? Because if you think about it, for the last 20, 30 years everybody's been talking about focus, core capabilities, stick to your knitting, really understand vertical expertise. And when you think about these universal capabilities around analytics data sciences, data processing, AI algorithm design and even algorithmic ethics, that are really general across any industry. We're talking about sort of general capabilities that can attack almost any industrial problem. You have companies like Zebra Medical Imaging, that do radiology, and use AI to detect anything from fractures to heart conditions.

Marco Iansiti:

And they have super minority of doctors in the company, 95% of the employees are not from the health care sector. They're a bunch of analysts, data processing people, computer science people, the founders are not from the healthcare environment. It's really interesting, and so from the perspective of the managers that are in existing established industries, it's a really scary thing. Because you have a company like Amazon or a company like AM Financial in China, that can come after your own vertical expertise, with a fundamentally different set of assets, different architecture and different set of skills sets in their organization.

Ben Newton:

Yeah, no I think that's really fascinating. I mean that's one of the reasons why this is so revolutionary. And one thing you guys mentioned too, because I want to definitely touch on a little bit how companies actually think about this, and some of the strategic planning you guys are talking about. But one more thing on the AI factory. You guys mentioned the idea of weak versus strong AI, with the context that, and I know I've seen this many times myself, is that there's a tendency to think that the solutions and kind of the approaches here are going to be super complicated. Is AI is going to go off and do some magic in a black box, and then something amazing happens. Whereas I think you guys do get at the fact that it's not always complicated, but it's always practical and provides value. So maybe talk a little bit about that. What do you guys bring that nuance out?

Marco Iansiti:

Again, a really interesting insight for us was that really basic AI can do very remarkable things. And that's why we get all excited about the fake Rembrandt at the beginning of the [crosstalk 00:25:09].

Ben Newton:

That was great.

Marco Iansiti:

It's like, you have a team of smart analysts and all that, doing the work of a genius, right?

Karim Lakhani:

Mere mortals [crosstalk 00:25:22].

Marco Iansiti:

It's on a team at Rembrandt, right? And a lot of people give us grief for that, because it's like, oh you're not really talking about AI, you're talking about some very basic things. And I'm like, but that's the whole point. The whole point is some very simple algorithms and do some very dramatic things. And once you replace even a simple process, that's got humans as bottlenecks with an algorithm, fundamentally the operating implications are completely different. And you don't need reinforcement learning to do this, you can take some pretty basic algorithms, recognizing edges to figure out the impact of radiology, to have a fundamental impact on the radiology field. And do the doctor's work.

Karim Lakhani:

Exactly, and also similarly we don't need to wait for the science fiction future, right? It's already here, right? Because what were see are these algorithms are actually very fragile, they do one thing and one thing only, but they do it really well. And compared to that particular human task, that AI was going to work really well. Playing Go playing chess as we've seen before. And now sort of recognizing images right, or recognized tumors. It does this one thing really well, and what's going on is that we can create a collection, and ensemble of these algorithms to do a bunch of things in sequence or in parallel, and it looks magical for us on the other side.

Karim Lakhani:

How is it that Google Photos know that I want to find these pictures of my taco, right? And recognize them from my image set, well those are a bunch of algorithms that are piled together. But they're all doing something very narrow, it's not a universal Star Trek computer, that's doing all these things. And I think realizing that this is both weak AI, it's narrow, but it's also achievable. You as a company can make this happen how, is what is sort of important for us to recognize internally, but then to also write about.

Marco Iansiti:

Yeah, and sort of thinking of AI as replacing a human being in some ways, it's wrong, right? In the sense that it's different. Another question we get asked a lot is, when is it going to be the AI is smarter than humans? And the answer is-

Karim Lakhani:

If I knew that.

Marco Iansiti:

How do you know you're not there yet. It's like if you go and you look at the Google search engine, right? Is that smarter than I am, it's like in many ways yes. Hell yes, it knows a lot of things that I don't know. But it's different, right? And so why should AI necessarily emulate a human way of thinking about problems? Maybe it's just going to be very different, it's going to be an ensemble of these dumb algorithms, relatively speaking. That can do remarkable things. And in some ways, things that individual humans or even organization's humans could not even come close to doing.

Karim Lakhani:

Yeah, yeah. And I think my favorite perspective on this is from Peter Domingo, who's a computer science professor at the University of Washington, who wrote this great book called The Master Algorithm. What he said is, people always ask him as a CS professor, when are AIs going to replace humans, are AIs going to replace managers or executives? And what he said was, AI is not going to replace managers, but mangers with AI are going to replace mangers without AI. I thought it was brilliant, that's exactly the perspective that we also subscribe to, which is this stuff is now so deep and so integral and so available, that those that can jump on this and figure it out for themselves and for the companies will have an advantage over those that don't.

Ben Newton:

Yeah, no I think you're right. And let's talk a little about that, because you give some ways to start approaching this in how you start making that transition, because I think another thing that I picked up reading the book is this is not just 100 year old companies versus 10 year old companies. This is also like you said, the culture. There are modern software based companies that are not there yet, they're still operating in what you might call a more traditional model, even if they're using modern technology. So how does a company at whatever stage start to tackle this? How do they start taking those steps in the right direction?

Marco Iansiti:

One of the things we emphasize in the book is that, virtually nobody has this right yet. And it's not about traditional firms versus modern firms, I think that there are certain things that traditional firms have taken very seriously, like for example privacy in many instances. Or cyber security and various other things that sometimes, the more modern firms haven't taken to heart in the same way. And they haven't built systems that are enabled to protect users and society and so on form misuse. And this is where we spent a lot of time thinking about the ethics of what we call digital scale, scope and learning, because when you're driving sort of information for a couple billion people, the ways to screw it up have enormous consequences, obviously. It makes the problem sort of all the more important and challenging at the level of society.

Marco Iansiti:

And so from that perspective in some ways, just like old style organizations have a lot to learn from new ones, some of the newer organizations have a bunch to learn from old ones as well. And so I think that over the next 10 years, I think we'll see a lot of organizations trying to develop a little bit of a hybrid model in a sense. Where sort of the kind of scalability scope impact and learning impact that you can have, is going to be in some ways moderated by the constraints set up by managing these things so they actually act properly. Around issues like bias, around issues like privacy et cetera.

Ben Newton:

Yeah, I mean that happens a lot, right? Because there's a lot of times this kind of underlying bias that came out of the development of this space, particularly some of the new technology companies, where... I mean Mark Zuckerberg has a few quotes attributed to him, which were problematic in terms of youth versus age and experience, right? But it does seem there's a transformation going on. Even some of these original companies like Google and Facebook and Twitter whatever, are coming face to face with the fact that there's some things that just haven't changed. You still have to have wise leaders.

Karim Lakhani:

Yeah, no and the wisdom is in some ways trickery. I mean I wrote a book years ago about ecosystems and platforms, and how... And I'll take the blame on this one, I was such an optimist. For thinking that ecosystems wonderful things, these computing platforms are just amazing things, because they enable millions of people to integrate and do great things. And there is a really positive story around that, but just in the same way the platforms could enable people to innovate and do cool stuff, the same platforms can enable people to do all kinds of terrible things, if they're used wrong. A used platform with data billions of people can be misused in all kinds of different ways, and some of the nefarious applications are coming to life over the last few years. And I think that's where the real thinking in some ways is happening right now. And you see a lot of computer scientists as well as the few of us in business schools, that are spending a lot of time thinking about how you design systems with the right ethics built in from the ground up.

Marco Iansiti:

Yeah, the big revelation for me and us, has been talking to our colleges in the computer science department at Harvard, because all of a sudden they're saying, "Oh yeah I'm studying computational fairness." I'm like what? I thought [inaudible 00:33:06] is the issue. Why are you as a computer scientist thinking about this in a mathematical way and an algorithmic way? And what they're realizing is that now, this ability for us to scale to 100s of millions of people actually has fairness implications, right? So for example, every human being is biased one way or the other. When you are in a traditional setting, your bias is limited to your local perspective, right? Whether you're a bank manger at a bank, and you discriminate against A, B or C, you just limit that discrimination to that one particular branch, and then you get found out and you get corrected and so forth.

Marco Iansiti:

But now if we use that data to train our algorithms, now we can discriminate at scale, right? Now we can do bias at scale, and so I think this notion that again this is both the engineering and the business and the philosophy coming together and taking this seriously, is a really important distinction for everybody to think about. And incorporate in how they design their algorithms, how they scale their algorithms, how they think about the products and services that are launched.

Ben Newton:

Yeah, that's a really interesting point, because we've had a few people on this program that have talked about the ethics side of that, and yeah there's always this ongoing friction. But there does seem to be a transition, that the way it's going to get solved is by the people actually writing the code, that are actually doing the work, starting to really truly take some responsibility for it.

Karim Lakhani:

Yeah, you need a multi faceted solution to this. I mean I think absolutely the core of this, you have to engineer it in from the ground up. But at the same time, I think it's also managerial responsibility, leadership responsibility, a board governance responsibility. I mean the board of Facebook right now is under the gun for insuring that the company complies with all kinds of different standards around privacy and end related issues. And so, I think we need to think about it form an engineering perspective, from a managerial perspective. It's really changing the game, we're managing algorithms instead of managing people more often then not these days. And so I think we need to understand how to view them.

Marco Iansiti:

I know a bunch of programmers, right, they're my friends. And they're not out there trying to be bias, right? They're a bunch of nice people. [crosstalk 00:35:41]. Most of them. But the point is, it's naivete, right? It's naivete, we didn't think that this is what we were doing. There's a sense of in psychology this unconscious bias, it's just there. So we don't want to get into a blame game, like oh you bad person you, this is what you did. I think we want to be like, hey, you didn't even know this what was happening, let's learn about it, and then let's be constructive about how we fix the problems. Instead of sort of saying, you evil person, this is what you intended to do from day one.

Ben Newton:

I think that makes a lot of sense, because there has I think we've kind of swung back and forth between that. And one thing that was interesting, a couple things I wanted to pock a little bit at, in this area, is you guys talked about two things that really stood out to me, or that I hadn't heard of before. One is that you brought up the story of Fidelity, and how they were making this transition and they build and AI center of excellence, which really reminds me a lot of, a lot of companies created a cloud center of excellence.

Ben Newton:

And it was this idea of kind of pulling this together, and really kind of getting a space where those decisions can be worked through. And you talk about information fiduciary, and this has come with some new ideas. So in that particular context, what mechanisms do you see, that the leaders can actually start to take? I mean one of this is they need to start thinking about this and taking it seriously and educating themselves. But also from a mechanism perspective, is it those kinds of things that are going to provide kind of a mechanism for starting to talk through this, or?

Marco Iansiti:

Our believe is that this has to be top down. The fast company change engines are going to get slaughtered, if they try to push this off themselves. You need the change agents inside the company, or we need top down support. And what we saw in Fidelity was from the CEO, Abby Johnson down taking a deep interest in this work, and committing to learning. Committing to surfacing the issues and making it happen. So we've been in sessions with some of their executives, where the entire C suite is there learning about reinforcement learning, and unsupervised and supervised learning, and all that kind of stuff. And I'm there taking notes, and they're trying to put it in the context of their business. Why? Well A you need to understand this stuff, because you're going to be asked to make decision. So you need to understand those decisions. But two, I think is that the signaling it provides to the rest of the organization, now this is important. This is where the C suite is thinking about is catalyzing. The [inaudible 00:38:22] is one thing.

Marco Iansiti:

We always shook our heads, trying to understand why did Sundar Pichai at Google do an AI first announcement? Saying Google is AI first now, when they were already AI first, right? They were already beating MIT, Stanford, Harvard in publications in AI, right? They were already creating the most patents in AI. Why is a company that's already leading in AI say, oh now we're AI first? And when we realized, this was a much as a message to internal Google people, hey guys we're going to be embedding this everywhere. So it was a message to customers, but importantly to the Google people saying, hey get on the bandwagon. Learn this stuff and figure out how to incorporate these things. So from Fidelity to Google, what we see is that the top down leadership is so, so important in making this happen. And then there are different structures you can create. A AI center of excellence. You can go to say, we're going to put a ton of resources. We're going to reorganize ourselves to be able to enable these kinds of technologies. Many options, but the top down leadership, super critical.

Ben Newton:

Well I guess in kind of putting a bow on this whole thing, because it's been a fascinating discussion. I really enjoyed the book, and I think what you guys are putting out here is very timely. Where does it go from here? So what are you guys doing coming up with this topic? I guess part of it-

Karim Lakhani:

We just finished the first book, you're asking us for the next book now? Come on.

Ben Newton:

Yeah, yeah what's your next book.

Karim Lakhani:

Oh my God. You're like my editor [crosstalk 00:39:59]. In all seriousness, kidding aside, look I think [inaudible 00:40:09] I'm just thinking, my sense is this book lays out a framework, right? And says what are the leading companies doing, that we can all learn from? And I think the question for the incumbents and the startups is, now what do we do, right? How do we now build ourselves to be in this way? And in many ways I think that's probably more important challenge, so the incumbent transformation that we see companies trying to adopt, right? You look at what Disney is trying to do in this space. Walmart is trying to do, and I think those are super important questions. You want to weep for Nokia, for not being able to make the transition, right? From a product company to a platform company. And the Scandinavian economy got impacted because Nokia went bust in the consumer business.

Karim Lakhani:

And we can't afford our best firms, the GEs the Boeing, the Fords the GMs, to disappear like that and cause all of these externalities, for their employees, but the rest of the economy as well. So I really take that, that's a big mandate in front of us.

Marco Iansiti:

I think it's such an exciting time right now, because I feel like there is a huge opportunity to study this right? So there's a major, major change that we've talked about. And when the model of the firm changes for us everything changes. So then there's a lot that we know so far about the engineering side of AI, there's much less that we know right now about the business side, the economic side of AI. And so we're spending a lot of time thinking about this. We're engaging with a bunch of firms, to see the challenges and opportunities from the ground up. We also have a laboratory here at Harvard, that Karim started about 10 years ago, that I'm now a co-investigator in.

Marco Iansiti:

Together with David Parks form computer science and [inaudible 00:42:12] form the med school. We're looking at a whole bunch of different research programs that we've ramped up, to try to understand a whole bunch of different basic challenges. Like what is the value of data? Data is the new oil, the new moat whatever you want, but how is it different, and how does it change, and how does it vary from one situation to the other? And what makes data valuable, and what is really the competitive advantage relating to that? So we can model that, we can measure that, we've done empirical experiments, we're doing more of that. We're doing impetigo projects to look at how advanced everyone is, to try to get a cut at where the world stands, with regards to information technology and AI. There's a lot of work to do, but it's an exciting time to go out there and get a sense for how people are doing, and what the leading lines are doing, and how everybody else is doing that's trying to catch up.

Karim Lakhani:

Yeah, and Ben can I make a selfish request?

Ben Newton:

Yes.

Karim Lakhani:

Okay, this is based on you said when is the next book coming out. So we learn by example and analogy, so your audience, your listeners are deep in these questions. And so we welcome them to reach out to us and give us examples, give us case studies, that we can study and learn from. Because it's at the leading edge, at the surface of I'm trying to do this transformation in my company, or I observed this company do something cool. We're so open to sort of input from your listeners, because that's how we learn, that's how we get better. And we're always looking to do cool cases on them. So [crosstalk 00:43:47].

Marco Iansiti:

Yeah, find us on LinkedIn or wherever yeah.

Karim Lakhani:

Exactly we're on LinkedIn, happy to connect. And then see if there's a fit in terms of what we're trying to do, and what they come up with as well.

Ben Newton:

That's great. I think you'll definitely get some takers on that, and again, I think the way you guys wrote this and the way you've approached this, it's very approachable. I think it's going to help a lot of people that have struggled with this really get their hands around it. And it's a great contribution to the space, and wish you guys all the luck. I think this is going to be great going forward, and we'll have to stay in touch. And everybody listening again, read the book. I don't know if say it was an easy read, because it's a lot of deep things, but it's a good read. I'm actually going to go read it again, because I want to really get some of these concepts, wrap my head around them myself. So Marco and Karim, thank you so much for coming on. I look forward to see what you guys do next.

Marco Iansiti:

Thank you Ben, it was a lot of fun.

Karim Lakhani:

Yeah, great conversation, thank you so much.

Ben Newton:

Absolutely. And thanks everybody for listening, and as always, look for the next episode in your feed. Rate and review us so that other people can find us, and thanks for listening.

Speaker 4:

Masters of Data is brought to you by Sumo Logic. Sumo Logic is a cloud native, machine data analytics platform. Delivering real time continuous intelligence as a service, to build run and secure modern applications. Sumo Logic, empower the people who power modern business. For more information go to Sumo Logic .com. For more on Masters of Data, go to Masters of Data .com and subscribe, and spread the word by rating us on Itunes or your favorite podcast app.

The guy behind the mic

Ben Newton

Ben Newton

Ben is a veteran of the IT Operations market, with a two decade career across large and small companies like Loudcloud, BladeLogic, Northrop Grumman, EDS, and BMC. Ben got to do DevOps before DevOps was cool, working with government agencies and major commercial brands to be more agile and move faster. More recently, Ben spent 5 years in product management at Sumo Logic, and is now running product marketing for Operations Analytics at Sumo Logic. His latest project, Masters of Data, has let him combine his love of podcasts and music with his love of good conversations.

More posts by Ben Newton.

Listen anytime, anywhere

Available to stream or download via these and other prodcast apps.