This picture is
before I got old. So before tenure,
you look young. And then you get very stressed. All right, so I'm really
excited to be here. It's the first time I'm
talking to, like, 1,000 people. So I'm a bit nervous. So I hope you don't
mind my nerves. What I want to do is walk
you through, for the first 45 minutes or so, some of
the insights that Marco and I've been developing
really for the past seven years at HBS. So seven years ago,
we started this course on digital innovation and
transformation, Digit. It's still in the EC. And we were both sort of
studying the software sector in different ways. And as we were
trying to think about what was happening in
the broader economy, the market structure, the
operations, the strategies that we saw emerge
in the software industry in the
'90s and the 2000s was spilling over into a
whole range of industries. And that started our
journey to rethink about, why is it that all industries
were looking similar, and why were there such
similar dynamics going on? And hence this deck
and these insights that we have generated. And really it was
the casework that we did talking to a whole
bunch of companies that got us to develop
these insights together. There is a ton of
hype around AI. It's probably the most
buzziest term around. If you don't attach AI to your
company name or to your CV, it's not going to be great. And so this is everywhere. I'm Canadian. So Canada believes it can
lead a global AI innovation. That's awesome-- to you know-- [AUDIENCE CHUCKLES] They are. Toronto has become the
hub for AI in many ways. Not the Raptors, though,
because I'm a C's fan. So a 19-year-old DevOp
code for the AI portrait-- America. But you can see this, right? Everybody-- there's a ton
of headlines around AI that are emerging. And this is not just
a theoretical thing, even for Harvard. So our cousins down the street,
MIT, is investing $1 billion in creating a new
college of computing. Thanks to a big donation
from one of our alums-- [LAUGHTER] --one of you guys. So it's like, wow, right? So this is like, you know,
Nithin is thinking about this, Larry is thinking about this,
Alan is thinking about it. Everybody's like,
wow, what's going on? And by the way, this is actually
amazing news for all of us. So if MIT is going to create
the supply of the technologists, those ideas, those
skills still need to be converted into businesses. And who best to
do that than here. So I actually see this as a
major feature and not a bug, that if we can understand
how to deploy AI then there's going to be tons of
opportunities for all of us here as well. So I think this is a
really good thing for us and not a bad thing for us. We still need to
raise $1 billion, but that's a different story. So of course, again,
around this hype, startups in AI, lots of
the number of companies that are like-- it was like
what happened a few years ago with blockchain. If you had a blockchain
to your company's name, you had a big boost
in investment. Same thing is happening
with AI now, too. Lots of companies are emerging. And whenever you see an
increase in volume of activity, like a like a loss
of entry going on, you should be thinking
about two things. One, people are
seeing opportunity. People are seeing opportunity. Lots of opportunities
are emerging. At the same time, probably the
cost of entry has also dropped. When you see a big boom
like this, people see gold, and I can get to the gold. And that's why we're seeing this
happen in the economy as well. But to ground this
and to get us to start thinking about this
in a broader way, I want to take us out
of the world of business and go to the world of art. All right, anybody take
art history as a major? Any guesses on who this
painter was for this? Rembrandt? Yeah, you guys are smart. Yeah. So this is known as "The
Next Rembrandt," which got created a few years ago. I'm gonna show you a
quick video about this, just so you can see the
same principles that go behind creating
the Rembrandt are also what is powering this big
revolution for us as well. [VIDEO PLAYBACK] [MUSIC PLAYING] - One of his great
achievements, one of Rembrandt's
great achievements, was to portray human emotions
in a much more convincing way than artists had before him
and, in many ways, for all time. - At ING, we believe in
the power of innovation, what it can mean to people. We want to bring this innovative
spirit to our sponsorship of Dutch art and culture. We knew that for
this challenge, we needed to team up with
experts from various fields to make this come to life. - We're using a lot of data
to improve business life, but we haven't been using
data that much in a way that touches the human soul. You could say that we
use technology and data like Rembrandt used his
paints and his brushes to create something new. - The first step was to
study the works of Rembrandt in order to create an
extensive database. We gather the data
from his collection of paintings from many
different sources, including 3D scans and upscale
images using a deep learning algorithm. Because a significant percentage
of Rembrandt paintings were portraits, we
analyzed the demography of the faces in these paintings,
looking at factors like gender, age, and hat direction. The data led us
to the conclusion that the subject should be a
portrait of a Caucasian male with facial hair, between
30 and 40 years old, in dark clothing with a
collar, wearing a hat, and facing to the right. From there, we started
to extract features only with faces that were related
to that specific profile. - And we had to create a
whole painting from just data. And we used statistical
analysis and various algorithms to extract the features that
make Rembrandt Rembrandt. - We took parts of the face,
and we started to compare them. And then based on this,
we were able to create a typical Rembrandt eye
or nose or mouth or ear. - After generating
the features, we were focusing on the
face proportions. We used an algorithm
that can detect over 60 points in a painting. We were able to align
the faces and to estimate the distance between the
eyes, the nose, and the mouth and the ears. - a painting is
not a 2D picture. It's 3D. You can see the canvas,
you can see the process, and that's what makes
the painting come alive. A height map is essential to
make the painting a painting. - We incorporate the height map
into the painting and print it on a 3D printer that uses a
special paint-based UV ink. It printed many layers,
one on top of the other, which resulted in the heightened
texture of the final painting. - It's sometimes
a magical moment to see a painting
for the first time. Even if it's computer generated,
for me it is something special. I would have believed
if I saw it in a museum that it would have been a
real Rembrandt, just one I haven't seen before. - It would be interesting to
see Rembrandt looking at it. He will be happy that there are
people trying to understand him and trying to create
something out of that. So I think he would be happy. - "The Next Rembrandt"
makes you think about where innovation can take us. What's next? [END VIDEO PLAYBACK] So what's amazing about
this picture to me was I was at an ad exec
session in our school. And unbeknownst to
me, in the audience was the head curator for
the National Portrait Gallery in Australia. And he was a Rembrandt expert. So I first showed this
photo, and he was like, it's a Rembrandt, and all
the things they said he said. And so for him-- I just have never
seen this one before, so it might be a
really good forgery. But what it showed was the
fact that this simple data can help you think about
the problem in a very different way. And non-experts in
art, the guys who you saw talking who had
no expertise in art, could then generate something
that looks like a Rembrandt. Now, of course, this is both
celebrated but also hated. And so these dual reactions,
these dual reactions are going to be
what many of us will be facing as we think
about the deployment of a digitized operation
in our companies that's going to
be powered by AI. On the one hand, we'll
be like, this is amazing, I love what this is doing. On the other hand, oh my
god, what about us humans? And I think that's going to be
the central debate that you all will be facing as you enter
back into the workforce and start thinking about these
ideas and their deployment. So one more foray
into the art world. This picture is the first
ever digital picture taken, by Russell Kirsch in 1957. And it was the first digital
image, 176 by 176 pixel. Of course, now your iPhone
has orders of magnitude more pixels than what this was. This cute baby is
Walden, Walden Kirsch. So this is the part where we
need to sort of understand the basics of what it means for
us to digitize the operations. When we digitize
something, what happens is that we can take this
and instantly replicate this at zero marginal cost. So this picture in '57
maybe taken in Boston, is instantly
replicable in Bombay and in Brisbane and Beijing
with zero marginal cost. So, the zero marginal
cost story is part of why digital operating
models are so scalable. The second thing is, today,
with some fancy deep learning convolutional neural nets and
all that kind of fancy jargon, we can take baby Walden
and try to make it better. And then we can also, you
know, maybe even age it a bit, so you have an
old-looking baby Walden. You've all done the
Facebook, you gave your face to the Russians, and
they age your face. You did that all, right? Something like
this was going on. The Russians now have
your photos, by the way, somewhere in Moscow or Belarus. So again-- so digital,
the data, the data can be used to
distribute widely, but also I can then apply
algos and make it better. And part of what we're sort
of seeing in front of us is that digitization has been
transforming our economy. And what I want you to
take away from this slide is the notion that
what's happened is that at each architecture
point in the tech industry, each technological
architecture in the tech industry, there's also been
an economic architecture that has also changed as well. So in mainframes and
corporate networks, we were all thinking about
product based companies, selling the magnesium grade
casings and the fancy lights. Then in the PC and web era,
networks start to explode, and the start of
platforms and ecosystems come through, where
companies competed not just on what they could do but
what their ecosystem could do as well. Then, in the era of
cloud and mobile, with Google and
Facebook, what happened is they said, oh, not only
do we compete with a platform and ecosystem, we can now
change the value creation, the value capture logics
that we have going on. I can give a lot of products
for free, I get the audience, and then I monetize
that with somebody else. So the architecture,
the technology changes, so does the economic
architecture, also changes as well. And today, in the world
of networks and AI, we're sort of in
this AI first world, where all these
industries are converging through these giant hubs of
Alibaba, Google, Facebook, Amazon, and we're now
thinking about the rise of these large hubs emerge,
where lots of companies now need to attach
through them to access the rest of the world. So this is going on. And this, by the way, changes
the market dynamics a bit as well. So I don't love this metric. It's like the market
capitalization per employee, but it's a metric that we
can all sort of agree upon that it has some meaning. And when you look
at a Walmart, which is a Ford, which is a Verizon,
which is Qualcomm, which is Goldman Sachs, which is
Ant Financial or Facebook, you see dramatic differences. So, like, Ford is working
pretty hard, right? Their factories are buzzing. They make billion dollar
bets on their factories on the next Ford truck. But the market is remorseless. Thanks for working hard, guys. I'm going to go to
Ant Financial instead. And so this is the economic
reality that we're faced with and we'll have to sort of
grapple with ourselves. And the point here
is that, by the way, Ford also needs the workers
that are at Facebook and Ant Financial, and so does Walmart. And so that becomes a big
challenge going forward for us. So I want to spend a few minutes
talking about Ant Financial. Those of you that have been
to China or are from China know how powerful a
company this has become. Ant Financial comes off
as a spin off of Alibaba. Alibaba was trying to basically
do e-commerce in China. The payments infrastructure
wasn't that great. There was no credit
cards, no credit ratings, that kind of stuff. So they set up Alipay
as an escrow system to simply allow payment
transparency and payment trust amongst merchants and buyers. Alipay sort of moved
along for a while. And then in 2013 or
'14, you get to see this super-linear scaling
of their user base. This is directly correlated with
smartphone penetration in China as well. But Alipay becomes
part of Ant Financial. Ant Financial doesn't just stop
at transactions and at escrow, but offers credit, banking,
financial cloud services, investment services,
you name it. By the way, the scale
here is that at the time there's a case on
this company, they had about 715 million
users, and that was powered by 10,000 employees. So 10,000 employees,
half technologists, powering 715 million users,
and about $1 trillion in transactions. So they're global. They're worth more
than Goldman Sachs. And they're
approaching valuations around Bank of America. But Bank of America serves
40 million customers, versus they are now approaching
1.2 billion customers. So again, a new type
of firm is emerging. Again, just think
about those that have been in the guts of a bank
or of a large scale company. How do you serve 1.2
billion customers? That's a big number. And Ant is a weird company. It's like, if you
look at their app, OK, there's payment and
scanning and pocket, but there are also
air and rail tickets. I can get city services. I can go to Taobao and purchase. I can do fan-- I can do donations. I can do all these
different things. So this notion that I'm
not just doing payments but I'm expanding into
your financial life is part of the story that we
see emerge along these giants. And they're trying
to basically-- any time you need money, they're
there, one way or the other, and they participate. And so the concepts of
a multi-sided platform have been discussed in
[INAUDIBLE],, of course. And so they try to connect
different types of platform participants in their network. And we know that
multi-sided platforms exhibit network effects. And network effects are simply
that the value of the product or service you're
using increases as the number of
users increases. Most products and services that
we have have a fixed value. Like, this remote control
has a fixed value. Even if all of us had
this remote control, the value of this
would be the same. What a network effect
says, no, the value of the product or
service will actually increase as the number
of users increases. So that's one big
thing around this. And this, by the way, of
course, is the same thing that happens with all the
other companies as well. They are all based on
trying to build network effects in their core business. Secondly, you can't
have human beings at the center of
this organization. It has to be robots,
it has to be machines. There's no way some
human beings are going to make a
decision on credit for a person at this scale. There's no way
some human being is going to make a
decision about fraud or not fraud in this system. Humans will be in the loop
somewhere, but not at the core. And the whole idea
here is that we have to use AI to
basically operationalize our bottlenecks-- remember our bottlenecks? Is that going to be on the exam. OK, he's not saying anything. I have no idea. And so that becomes
part of what allows them to scale in this way. So let's just get our
definitions right. What is AI? So this is the theory
and development of computer systems able
to perform tasks normally requiring human intelligence,
such as visual perception, speech recognition, decision
making, and translation between languages. And there's lots
of drivers for AI, forces that more and more
data is being generated. Everything we do
now is data-fied. Any activity you're
doing is being data-fied. Right now, Google and
Facebook and Amazon and Instagram and
Snap, they all know you're here at this location. You're broadcasting your
location to these platforms. There's lots of
computing available that allows us to
analyze this data. And there have also been
great advances in algorithms. So we should just get clear
about the definition of AI. The first definition
of AI, which I love, machines that can
think and act in a way that it matches or surpasses
human intelligence, is known as strong AI. This is the AI of science
fiction, of Star Trek. And this is where we get scared. We go, oh my god, strong
AI is going to come, and robots are going
to take over the world. Another definition is weak AI-- any activity computers are able
to perform that humans once performed. That's weak AI. And what our thinking is,
what Mark and I are thinking, is that this actually is what
you need to get the ball going. Google and Facebook, or all
the multi-sided AI startups don't have strong AI. They take one or two or three
or 100 activities and say, can I do it better
with algorithms? Can this become better or
not-- and just deploy that. And there's lots of
use cases of weak AI replacing or augmenting humans. Let's take this Uber Eats
example one step further. So the data-- what's
going on with Ant? What's Ant doing? Ant's taking the data they have
from one set of transactions and saying, I can now spin
up and go to the next set of transactions. They're expanding scope. Scope expansion is part of what
we see many of these companies do. And in many ways, this
runs counter to the advice that American management gurus
gave in the '70s and '80s and '90s. Tom Peters, In
Search of Excellence, stick to your knitting. Do one thing, and do
one thing really well. You need to do that for sure. But these guys are
saying, let's actually think about scope expansion
because we have the data. And that data allows us to
cut across a range of markets. The other thing
is that they also use the data to drive
digital learning. The whole setup
within a platform is, as you get more data,
you build better algorithms. As you get better algorithms,
you get better service. As you get better service,
you get better usage. You get more data. And this flywheel turns. And that's why the data
side is so important as much as the algorithmic side. And what we are sort
of talking about is that now, many companies
will need an AI factory inside of their operations. This is not like
Donner, thankfully. And this factory,
by the way, is going to be the same regardless of
the company you're running. So the McDonald's AI
factory will look the same as the Ford AI factory. A hamburger factory in reality
is very different from a car factory, but the AI factory
is going to be the same. And the elements here
include a data pipeline, algorithmic development,
infrastructure development, and experimentation
platform, and your ability to prioritize and deploy. This becomes the core
aspect of the operations that many companies will
now be striving for. So let me give you an
example from China again. Pig farming in China-- this
is a great awesome example about how you get data. So JD Digits, it's a
stepchild of jd.com, trying to do in finance. China is the largest pork
producer and consumer, occupying 56% of the total
production in the world, market size, close to 200
billion US dollars per year. But more than half of China's
nearly 700 million pigs are raised on small-scale
farms, family farms with less than 500 pigs. And the Chinese
government is encouraging these large companies to
provide financial security. So JD Digits thought, oh, fine. Let's offer insurance on
the health of these pigs. So the problem,
though, is this is going to be ripe for
fraud, because how do I know which pig got sick? It's a real problem, right? Like, if you're going
to build an insurance product around
insuring livestock, you must be able to
identify them one way or the other in an easy way. Unless you sort
of did a DNA test and sent it to the lab
and that kind of stuff, it's going to be very difficult. So these guys came up with
pig facial recognition. Let's data-fy the farm. Let's data-fy the
farm, and then use that as a way to build a
financial services product that I can go to market with. So let's have a look at this. So there's lack of
label pig face data, similar to what Amy was saying. There's lack of label data
around the medical records. So they basically
went to some farms, put some cameras on, and started
to build an algorithm for pig facial recognition. And CNN is a
Convolutional Neural Net that allows you to do that. And so this allows
them to now start to create products
in interesting ways for the farmer. So not only are they doing this
and have an insurance product, they can then also
say, hey, I've got this digital
infrastructure now in the farm. I can now give you
weight estimation, I can give you how
healthy the pig is, I can tell you how
you should feed them, and so on and so forth. Think again scope expansion-- I can now be in the supply
business for farmers because I know what's
happening with their livestock. And now you might
be wondering, well, how will they know in a
moving farm what to do? Well, they also have a
solution for that, too. So they have got tools
to count pigs and know which pig is which and
what they're eating and what they're not
eating along the way. This is data-fication,
the data pipeline story, happening within a
farm setting, and then using that data not to sell
the technology to the farmer but actually to create services
on top, financial products services on top of that. So data-fication becomes a
key element to think about. The second thing is, we talked
about algorithmic development. What's nice is that
much of the algorithms are available off the shelf. So the cloud providers
like Google, like Amazon, like Microsoft, have made these
solutions available for you as plug and play. Again, you still need a
data science expertise, you still need to be able
to connect them together, but no longer are we spending
$40, $50, $60 million trying to create the algorithms. Those can be gotten
off the shelf. And the zebra imaging case
gave you a hint of that. They were spending a lot of
money on algo development, and then they shifted over
to TensorFlow from Google. So this is now
becoming off the shelf. And it's so off the
shelf in some ways that even a faculty member
at Harvard Business School can now write papers about AI
in cancer detection and lung cancer segmentation and get
published in the top oncology journal. So this is me-- the last time I took biology
was in the 10th grade, and I got a C. And now I'm
publishing in a top life sciences journal because
we were able very cheaply, using crowdsourcing, to
generate these algorithms and actually now be as good
as the average radiation oncologist at Harvard
Medical School. We spent 80 grand and
eight weeks in prize money to get this done. So again, what I
want to tell you is that these tools are
becoming widely available and are relatively cheap. So let's think about what
an AI factory looks like. This is one AI
factory at a company called Ocado out of the UK. They are building
warehouse automation for food for grocery stores. And this is a fully
automated warehouse. These robots go on on their
own, do their own thing, pack, pick, deliver,
make it all happen. Ocado now has the
ability to give you the providence of your cheese
that you got from the farmer in the Midlands,
that level of detail. And now, given
all the data, they can now predict what you
will want to order basically four days ahead. So they can actually
be ready and packed even before you put the
order in because they have all the data that you have. After a while, once you get
enough data about your shopping habits, we're pretty
predictable people in terms of what we want to
get from our grocery stores. So here's an example
of a working-- this is the back end
of an AI factory. Again, the same
approach that Sam had put into place
for her ad product is also running
behind here as well. And then here's the AI
factory from Amazon. [VIDEO PLAYBACK] - Four years ago, we
started to wonder-- what would shopping look like
if you could walk into a store, grab what you want, and just go? What if we could weave
the most advanced machine learning, computer
vision, and AI into the very fabric
of a store so you never have to wait in line? No lines, no checkouts,
no registers. Welcome to Amazon Go. Use the Amazon Go app
to enter, then put away your phone and start shopping. It's really that simple. Take whatever you like. Anything you pick
up is automatically added to your virtual cart. If you change your mind about
that cupcake, just put it back. Out technology will update your
virtual cart automatically. So how does it work? We used computer
vision, deep learning algorithms, and sensor
fusion, much like you'd find in self-driving cars. We call it Just
Walk Out technology. Once you've got everything
you want, you can just go. When you leave, our
Just Walk Out technology adds up your virtual cards and
charges your Amazon account. Your receipt is sent
straight to the app, and you can keep going. [END VIDEO PLAYBACK] So again, the AI factory. That's going to be
a core component of a digital operating model. So we've talked about
it in the course here about business models
and operating models. Business models is about
how you create value, how you capture value. In a world of
platforms, you also have to think about
value sharing. But then the operating model
is about how you achieve scale, scope, and learning. And so if we go back to why
this is so different now-- scale, scope, and learning
has always been around. A book called Scale and
Scope by Al Chandler won a Pulitzer Prize. Al Chandler was a faculty
member here at HBS. So scale is about how you just
get more and more customers. How do you serve more and
more customers efficiently? Ford taught us that. Scope-- have more variety. Sears taught us
that before Amazon. And Toyota, TPS, taught us
about learning and continual improvement. These images of these
companies, by the way, don't look that different today. They all look
about the same now. And what happens inside of
these companies is that in order for us to achieve scale,
scope, and learning, we become into silos. We set ourselves up into
silos, where the IT group talks to one product team and
only one product team, and it's all that they work on. And the same thing
across the way. So the data that we have
is wildly fragmented and not available for
us to drive the insights that we're talking about here. We were chatting with
the CIO of Goldman Sachs, and he said there are 30,000
employees at Goldman Sachs, and there's 95,000 databases. Why? Well, it was very efficient. At Goldman Sachs, you
were just out there to go get that deal done. Figure out that business
line and make it happen, and don't be constrained. And that works really
well in that world model. But in a model where
the data is important, data of all consumers at
all times is more important, we have to rethink how
we organize ourselves. So what happens in this
traditional operating model is that the
value of the firm is constrained, because over
time, as the number of people we serve, the
demands that we face in terms of complexity, cost,
organization inertia, basically plateau out the value
that we generate. That's the reality that many of
us face in large organizations. Getting anything done is
difficult, it takes a lot time. If I want to get data from one
setting to the other setting, it's very difficult. I face this here at HBS. We have an MBA product, we
have an exec ed product, we have an online product,
we have an ER product, we have a publishing product,
and there are nine different IT shops within HBS. And they don't share data. So if you are also
reading HBR, I have no clue what
articles you're reading, even though they probably have
that data available to them. Or if somebody goes to exec
ed after five years at HBS, we won't be able
to track, oh, what classes they take, what
professors they saw, and so on and so forth. Normal things that
you'd expect, we also face a problem here as well. And so digital operating models
have these zero margin costs and can generate both
these network effects but also learning
effects, because what happens is we get increasing
value as our platform grows, we get more and more data, AI
comes in and basically sharpens this curve for us. And so the traditional operating
model performance drivers are such that we have to sort
of think about scale, scope, and learning, but in a world
where we have digital operating models, we now think
about scale in terms of zero marginal costs. We think about scope in terms
of aggregation and modularity across networks and learning
in terms of constant innovation and AI and ML. And so what happens-- this is a traditional
product business. This is now faced
with the collision from the digital business. You have a decreasing
returns business with an increasing returns
business colliding, and this collision happens. And we see this collision
happen over and over again. Nokia versus Apple,
Marriott versus Airbnb, Ford versus Waymo, HSBC
versus Ant Financial. And by the way, this
debate is live at HBS. We have the MBA
product, but in order for us to double the
capacity of the MBA product, we'll need to
build two Klarmans, take a long time, and more
donations from people like you, and a few decades. We'll have to double our
faculty size and expand space. But if I go online, I can
just scale in a massive way. And this creates a real
puzzle even at HBS, because the MBA product looks
at the digital product and says, hey, what's going on? I've got way better
value than you do. You're like a sink
of money for us. At the moment, we are investing
a ton in the digital product. And this is what Harvard
has to figure out as well. We have a traditional
product business, and we need to think
about an online business as well, because our
mission for education is not just people here
but the whole world. And so this is
creating attention within even our school. I think we are in
this transformation stage for the economy,
and not just one sector. Like, in the '90s, I got
my masters in the '90s, and it was the tech sector
that was transforming, and lots of the bubble
was being set up for the internet at that time. And the bubble bursted, and
then we came out of it stronger eventually. But now almost all sectors
are facing the same types of transformation
opportunities, and I think that's where you guys
have Greenfield available to you to go forward. So let's give the bigger
picture and sort of set this in terms of new rules. Star Wars is coming, and so
there's a question about, is it the Empire or
the Rebel Alliance that's going to win in the end? We'll figure it out. All right, the first is-- this should not be a
surprise to you anymore, digital technology
is everywhere. It's being embedded into farms
and into your back pockets. What's happening are that three
specific laws are converging. Moore's Law, which
says I can get faster and faster computation
at lower and lower costs. Metcalf's Law, which says
connectivity and value increases as I build networks. And Barabasi's Law,
which says that when you go into a world of networks
and platforms, hubs emerge. Hubs emerge through a mechanism
called preferential attachment, which drives the creation of
these large organizations. So this is happening
across the economy. And this has
enormous implication for the ways in
which we regulate, the ways in which
we run companies. Not all companies will
be platform companies. Most of us will be working
in non-platform companies. And so we have to figure out
how we compete and participate in their ecosystems. And that's going to be a
key challenge that we face. Second is that there's
turbulence, right-- all of this constant improvement
in the technology drives a ton of uncertainty going through. And there's lots of examples
of people feeling unmoored along the way. The third thing is universality. Just as I was discussing that
the McDonald's AI factory is the same as the Ford AI factory
as the Amazon AI factory means that we have
to now be thinking about data and analytics
as the core drivers of what the firm does. We'll be living in this
world of digital collisions, and we have to figure out
how do we organize ourselves, and what skills do we build,
and what skills of leadership that we have to go across
these organizations. So in order for you to run
a modern McDonald's, you might hire somebody from
Ford, which would never have been the case
beforehand, because there was no universality
of those verticals. But now we sort of see
this universality emerge. The fourth element
here is recombination. We're seeing companies
and the boundaries of what companies are being redefined. There used to be a
telco sector and there used to be a banking
sector and there used to be a social media industry. They've all merged together. So this recombination
because of connectivity, again, opens up
new opportunities, but also creates new challenges. We see these hubs
emerge everywhere, and then how do we participate
in these hubs becomes important. And then the most
valuable public companies are basically the ones that
are creating these hubs. I don't think in
the long term we can sustain this inequality
even in company profitability or company valuations. And that'll be part of the
challenge that we all face. This ethics story
is very important. We need to engineer
the ethics now. We have to think
about ethics at scale. So human beings are biased. We're racist, we're
sexist, we're this or that-- that's just
part of who we are. We're flawed creatures. In the analog world,
you could basically have your bias limited. In a digital world,
you can bias at scale. I can train unbiased people,
I can train unbiased data, and then I can discriminate
on a highly scalable way. And so these issues around
selective amplification, bias, control, privacy,
these are going to be top, top issues that
business people will face. It's no longer a
question that we're going to let our law school
cousins deal with this or the Kennedy School
cousins deal with this, because this will be
a boardroom issue. This will be a boardroom
issue that you will all face as you design these
systems going forward. So questions around
cybersecurity, transparency, concentration,
and so forth, comes out as well. So here's a call to action. You have to sort of understand
and actively anticipate the transformation of our
social and economic environment. We have to drive this
innovation and operating model transformation to create
this foundation for change. We have to focus on our
own digital business models and leverage these new digital
opportunities for new revenue, collaborate to break up
the competitive bottlenecks because they are
going to be there, support a
multi-platform economy, drive interoperability,
multi-homing-- those are going to be
key strategic but also technological issues
that go hand-in-hand-- leverage partnerships,
communities, and crowds to drive alternative
platforms, and then also understand the
regulatory options. Regulators are in business in
Europe and China and the US, and we have to figure this out. So a list of further readings-- if you care about platforms,
the first three books, Second Machine Age,
Platform Revolution, Business of Platforms,
recommend highly to get into the
economics of platforms. Ming Zeng's book
on Alibaba, Smart Business, awesome
book if you want to really understand the rise
of Alibaba and Ant Financial. AI Superpowers, a
political economy view of China versus the US--
again, a great book. I really love
Prediction Machines. This is a book that really
talks about the economics of AI and the economics of
prediction as one part of AI. Marco and I have a book coming
out, Competing in the Age of AI, in January. And then I also love
science fiction. So William Gibson, Ian Banks,
Ann Leckie, great science fiction writers who are
thinking about machines and AIs and humans. William Gibson thinks
maybe 30 years ahead. Ian Banks thinks
centuries ahead. And Ann Leckie has this great
Ancillary Justice series where the protagonist is the AI. The protagonist is the AI. So as a thanks to Elise
and Sam and to Avery, we're going to give them
a copy of the book now. So thank you very much. But all of you get a
copy as well outside. So we were able to-- the
book comes out on January 7, but out in the hallway are
the copies of the book. Thank you so much
for taking the time. [APPLAUSE]
