So, Kareem, it's going to be exciting to hear you talk about this incredible hyper-

speed
with which AI is all around us.
We had this enormous kerfluffle a few weeks ago with Chad GPT, and everybody is
around
there playing with Dali and Chad GPT and Redjourney and what have you, but I'm
hoping
you can give our participants some sense of the evolution of how we got to this
point
and why it's relevant for every single person on the planet, including, you know,
people
in slums in India, people across Africa, the middle of the United States, China,
wherever,
anywhere, whatever sector they're in.
And I'm just conscious that we're sitting here on this project for Aspire Institute
and we've
tried to, because of your orientation primarily, and then laterally mine as techno
files, kind
of built this organization as a digital first-world position.
So I'm excited to hear what you have to say.
Yeah.
Yeah.
Look, it's easy, you know, for people to say, well, this isn't a part of me, right?
Like this digital stuff was only for America or China or fancy stuff, and it's
going to
be many fancy examples.
But I think the key key thing for our participants is to say, we're already living
in a digital
age.
You know, these phones that all of us have, these smartphones that all of us have,
are
supercomputers, right?
And they have GPS, they have all this key technology already in your pockets.
And so the fact that most of you are seeing this digital leak, right, through our
digital
platforms, the best thing, we're not forcing you to come to Harvard or to invite
you,
but many of us, you know, we can't afford for all of you to come to Harvard,
because
that'd be very expensive, but we can get to you.
And so in many ways, your personal lives are already imbued with digital
technologies.
Absolutely.
And so as you start to think about how you're going to drive change in the world,
you want
to actually have, seriously about how digital becomes core about what you do.
Both how it's an enabler of things and how you can use it as a consumer of things.
Exactly.
And I think, you know, and just, just, you know, in our reflection when we started
to
build a spire, what we quickly learned is that our intentions for impact, our
intentions
to reach more people, us not being just cute, but actually having a war-light
impact and
having a skill really can only be enabled through digital technologies and through
AI.
So keep in mind, I'm going to show you lots of kind of crazy examples, but keep
thinking
about the fundamentals, that's what we're going to talk about, and how those
fundamentals
apply to you and how in your endeavors, you start to think about how these
technologies
can be helpful for you.
So not to belabor it, but I came across a statistic that I've since been validating
that the country with the greatest per capita use of data today is India.
It's India.
It's a poor country.
Yeah.
Right.
So if you add up the per capita use of data in the US and China, India exceeds it
by
40%.
My goodness.
And the reason for that is that there is a digital transmission of the country
scale
that's right.
That's right.
So I propose your point that this is applicable to everybody.
Yes.
Yes.
It's just in a corner store right next to you.
Exactly.
Exactly.
Great.
So let's go try and make sense of this thing.
So 10 plus years ago, Google paid a lot of money.
We were talking about Google in the previous session, but Google played a lot of
money
for Nest.
Yes.
Three billion plus something like that.
Let's talk about that.
I know you've looked at that.
Yes.
And what is the reason for that?
Yeah.
Yeah.
It seems like a lot of money to be.
Yeah.
Yeah.
So look, Nest made a thermostat.
So for those of you that don't know what a thermostat is, thermostat basically
allows you
to control the temperature, up or down in your house, in your office, in your
school and
so forth.
And you know, the headlines were kind of interesting like, oh, it's a distraction.
And you know, all the fun tweets that came out of that as well, you know, as like,
you
know, all of a sudden Honeywell, which is a big traditional manufacturer of
controls,
like thermostat is like, oh my God, I'm competing against a search engine, right?
And so on and so forth.
And some people, that was a big distraction.
So let's go watch a video first about what Nest is.
So you understand what a digital thermostat actually does.
This is Nest thermostat from Google.
It's a smart thermostat that can help you save energy, which helps you save money
on
your energy bill.
And with all those heating and cooling savings, replacing your old thermostat is
almost a no
brainer.
You're probably thinking, oh, wow, but how?
Well, this thermostat can help you build an energy saving schedule and then adjust
automatically
so you don't waste energy when you're away.
You just have to select preferences such as what temperature you like but when
you're
home, or when you're asleep, or when you're gone.
And then plug in your schedule.
You know what else is cool?
The Nest thermostat is kind of like an energy saving advisor.
It can help you find opportunities to use less energy.
Like it might send a suggestion to tweak your sleep temperature settings.
So you can save while you snooze.
It can also help keep tabs on your heating and cooling system and sends you helpful
reminders.
Like when you need a filter changed.
It can even alert you if something doesn't seem right and help you find a qualified
pro for a system checkup or repair so you can get things working right and go back
to
living life.
And see you're at the park and want to come home to a perfectly cool house.
You can go to the Google Home app and take care of that.
You can also control the thermostat from your Nest speaker.
Hey Google, raise the temperature.
Maybe you're thinking, is this fancy thermostat real expensive?
Well, actually, it isn't.
With rebates from your utility company, you can get it for even less.
Oh, and if you like to DIY around your HRME.
The Nest thermostat is designed so you can install it yourself.
So you don't have to hire anyone to do it for you.
But if you want, you can have a pro come install it instead.
And that's how you can save energy and money.
With Nest thermostat from Google.
So let's first before we actually dive down into the business reasons
about why Nest was so valuable.
Let's first understand digital in the first place, right?
Because that's a core thing.
So we're going to do a little bit of digital 101, I mean, it can be five properties
and how they actually enable our digital world to exist.
But also then what that means for business and economics.
So the first thing is that digitalization is all about taking an analog signal,
and quantizing it so that you can make it at one or zero.
Digital is all like binary one and zero.
And one is a one and zero is more efficient for us to transmit it, right?
And once you once you have it in a one and zero form is actually quite it becomes
error free in its transmission, right?
Because then you can just transmit the ones and zeros.
You don't have to transmit the entire analog signal.
You can just take a sample of it and ship it through.
So it tends to be entirely error free in its transmission.
The other three properties are that once you have something digitized,
you can keep recreating it over and over again, right?
It doesn't cost you any extra.
Yeah, exactly.
The marginal cost to reproduce it is relatively zero.
And because we need to now transmit the signal around the world, right,
there's a lot of effort been to standardize the language of digital communications.
So, you know, we basically have Esperanto.
Esperanto is, you know, the universal language of how our computers communicate.
And then finally, and we discussed this a bit earlier,
is that the transmission costs are relatively low, right?
There's near zero marginal costs for us to be able to transmit these signals.
Yeah, so just to clarify that, the fact that I send this signal
that you doesn't degrade my ability to send it to somebody else.
That's one thing, right?
So that'd be like the same.
So when you watch a web page from Harvard, it's coming out of Cambridge.
But it's exactly replicated somewhere else in Bangalore, in Boston, you know,
you name it as exactly or Beijing, right, exactly replicated, right?
And so that's like this error-free transmission,
but also it's infinitely reproducible.
We can have a web page coming as many times as we want, right?
And it doesn't cost Harvard any extra money for another reader in Beijing, right?
Yeah, it's causing a server to crunch and pollute somewhere else.
But that's a different story.
Yeah, but for Harvard, it doesn't cost the same.
It doesn't cost the same.
Exactly.
The person has to pay to get access to it, but Harvard, as a producer, does not
have to pay.
That's great.
So once you understand that that's the core property of digital signals,
that allows us to say, well, what can we do with this?
What does this mean for business, right?
And so when we up to these, these, these, these, these properties are going to help
us
understand how different people have manipulated and responded to digital signals.
Exactly.
And we'll, now we'll bring it back to Nest, right?
Because Nest was a digital thermostat, right?
And what made Nest so valuable?
So, so, do you remember when you first came from India to, to the US?
I would call it.
I would call it here.
Yes, yes.
Right?
And like, you encountered your first ever thermostat?
Well, actually we had no thermostats.
We had furnaces.
Yes.
In, in my Princeton dorm room.
Yes.
And it made loud noises like this.
T-t-t-t-t-t-t-t-t-t-t-t-t-t-tlight.
Yes.
Right.
As the oil and the heat kind of went through.
So, we didn't have any thermostats.
In America.
In America.
But you came like in...
In the dark ages.
In the dark ages.
Yes.
Before I was bald.
Okay.
And so, look.
So, let's think about what a thermostat does.
Right?
So, like in the northern climates, right?
You have an analog thermostat.
Yeah.
You know, you install a furnace, right?
You, you install a thermostat to regulate the temperature.
Yeah.
You buy fuel, right?
And then you set the temperature in the thermostat.
And then as the thermostat sort of senses the changes in temperature, right?
It will turn the furnace on and off.
Right?
That's all it does.
And the thermostats have been around for 200 years.
Yeah.
Right?
They use mercury in them in the old days and so forth.
That was sensitive to temperature change.
And it would, you know, trigger some electrical circuit to turn the furnace on and
off.
Now, what's interesting is that the digital thermostat does the same thing, right?
You set up to buy a furnace.
You set up to buy a thermostat.
But you set up to buy a fuel, set the temperature, sense the temperature and turn
the furnace on and off.
But now what's different, so it's the same process.
But now you actually have data because you digitize things and you quantize things,
right?
You can now have data on buying fuel, on setting the temperature, sensing the
temperature and turning the furnace on and off.
So it's the same functions going on, but now you have data.
And once you have this data, what it allows you to do is to have way more control
and intelligence.
Like what we saw happen in that video, right?
You can now change the temperature and you're in the world, right?
You can now have a preference for this temperature and say, well, you know, if the
electricity bill is going to be too high, right, the cost that is going to be too
high, don't turn on the thermostat, the furnace or the air conditioner and so
forth.
So you have local intentions and control.
But the interesting thing is the following.
Then once you have data from one thermostat, what can you do?
You can aggregate it with data from a bunch of levels.
All thermostats, right?
And so that's where we start to see why the value of a nest became so interesting.
Because now what you can do is I provide you with a lot of value for your local
thermostat.
You pay me lots of money because the nest thermostat was like $250 at the time it
came out versus the average thermostat was $25, the conventional thermostat was
$25.
But now with this cloud infrastructure, you can now basically start to connect all
the thermostats.
And then for a company like nest, you can do all this advance analytics.
And oh, now that I have data across a municipality of thermostat and temperature, I
can now advise energy companies on what to do with the thermostat.
I can advise consumers on how they maintain.
So an example that I can think of in my backyard in Boston is you frequently see
your consumption.
And you get a summary statistic of other people's consumption in your neighborhood.
And that might prompt you to change your behavior.
Exactly.
And now you can do this dynamically.
You can do this dynamically.
Once you have a digital thermostat, you can also connect it, for example, to your
firearm.
Because what happens when a fire alarm, if a fire happens in a house, typically the
AC starts up because the temperature has gone up.
That blows air into the fire, which is actually really bad.
And so with a smart thermostat, what you can do is say, oh, there's a fire.
I need to actually, so I can have a smart smoke detector, which says a fire is
happening, shut down the thermostat, shut down the AC and not have the fire in
control.
And then you can even do things like trigger maintenance.
And so what Google saw in Nest was this powerful tool inside the home.
That was going to basically give them an entry into a smart home business.
But it's based on this core foundations.
Let me ask you, why did Google see this?
What allowed them to get a jump on this?
Why Honeywell didn't do this?
Or anybody else?
Or other Czech companies?
I mean, they paid three billions.
They must have thought this was a lot more.
And they saw it.
Because if everybody had seen it, the price of that asset would have been bit up
quite a bit more than three billion.
Absolutely.
I think, A, they were like a data first company.
They were used.
They were like a spire.
They were like a spire, exactly.
Oh, we'll be at Google's eyes soon.
That's why I appreciate it.
Because they were in this world of data, they were all about information.
Google's mission was to aggregate information from around the world.
They go, well, the thermostat is another source of valuable information.
It gets me into the household.
And they were thinking about the fact that they were not just a search engine,
but were going to be available to anybody anytime, anywhere.
And so that's how they saw the value and that.
And also importantly, they could imagine not just the value of the sun, the device,
but then the ecosystem underneath it as well to make it work.
So it's so interesting.
It goes back to your opening point, right?
That this applies to any setting.
I would imagine that Honeywell is sitting in Minneapolis, so wherever they
headquartered,
didn't imagine this because they didn't have that cognitive orientation at that
time.
They were just selling devices.
They were selling devices.
They were not focused on data, right?
Because once you focus on data as your core engine, you think about it very
differently.
You see value of data in many different ways.
And so then what we are in this world of is this transition.
The world is transitioning from analog systems to digital systems.
That's been going on for the last 20, 30 years.
But this is what's going on.
Now you went from this old thermostat to now a thermostat that's fully digital,
fully controllable through your smartphone and so forth, Wi-Fi connected and so
forth, right?
And then people always ask me, like, how would Google make money?
So a student of mine said me this cartoon, right?
Well, Google's whole thing is all about advertising, right?
So when the Nest smoke alarm goes off, right, Google's going to send you
advertising for fire extinguishers
and hotel rooms and so forth, right?
It's temporary housing, right, until you recover from the fire, right?
Because Google's businesses aren't in that business.
But they can monetize that data for additional things.
So that's a terrific example.
Because it's an everyday device occurrence.
As you point out, we've been, we worry about all humans.
You know, you asked me about what happened when I came to America.
But let me tell you what happened when I was a kid.
And we lived in New Delhi, which is extremely hot in the summer and quite cool in
the winter.
But our thermostat was, we had a three-story house, including the basement and the
two stories.
And the thermostat was when it's really hard to go to the basement.
Yes.
It's really cool.
And it's really cool that you can come up during the day.
Yeah, yeah.
So this is, you know, temperature modulation has been a part of human life forever.
Exactly.
That's, you know, we are warm-blooded creatures.
Yeah.
We need to have that.
And we're just bringing that forward, just like we saw Alibaba, the other example,
bringing the idea of haggling and bargaining into the online era.
Exactly.
So at some level, there is nothing new.
There's nothing new, exactly.
And what's interesting, though, is that you can generate a ton of value as you say,
how will I do this whole thing in this new way with digital?
That's right.
That's great.
That's well said.
Exactly.
So let's just take a step back and just think about the fact that worldwide,
especially
in developed markets, there has been a relentless investment in technology and IT.
And you know, over the many decades, we've gone from less than 10% of capital
expenditures,
so like what happens, how companies allocate money going towards digital to now
approaching
close to 25%.
That's a huge change.
It's a huge change.
And in that era, what we've seen is all these giants come about, right?
Like IBM and Microsoft and SAP and Netscape and Amazon and Google and Salesforce,
all these
companies came about because the whole world set to each other, the business world
set
to each other, that as more and more computers become available, as the cost of
computation
goes down, right, we can do more things with this and more opportunities show up.
And I think that's a really important thing because this trend is going to keep
continuing,
right?
More and more of the world is going to get digitized.
More of the ways in which we interact is going to get digitized.
And then the key for us is to say, what can I do with it?
What are the opportunities that come up from this to make it happen?
So think about some of our participants, right?
Whether it's participants in a Harvard classroom, whether it's the Aspire Institute
participants
who are living their lives, you know, buying and selling fish in the Karachi fish
market
that you were talking about earlier, or as part of a microfinance community in
Mexico City.
Yes.
Give me some sense of how this explosion of data is going to matter to any of those
settings
or pick any setting.
No, 100%.
So look, like let's talk about the fishermen, there's been some great examples,
right?
So a few things.
Now with the smartphone, now you probably can get meteorological data, right?
So you know, like when storms are becoming or where the temperature of the water is
going
to be and where the fish are going to be, right?
That's before it was like instinct, it was sort of smell of the air, wisdom, like
red
sky, sailors, like, you know, all that kind of stuff, like, you know, so those are
all
wisdoms.
Now you can quantify it and you can have it available to you in the smartphone, so
it makes
it like easier.
There's these great examples of when fishermen go out there and they're fishing,
they need
to land at a port to sell their fish, right?
And different ports and different brokers will offer different prices, right?
And so now they can get immediate access to saying, where should I go?
Which port will offer me the best price for my fish?
Where is their relative scarcity of fish that I go there?
Exactly, exactly, right?
And so before you would always go to the same port and maybe your broker cheated
you because
the supply was a lot and then you were like, okay, the price is going to drop,
right?
But now you know exactly where the demand is going to go and you can go that.
Or even agriculture, right?
Same with agriculture, right?
As we were talking earlier, a big portion of the world relies on agriculture as a
source
of employment and so forth.
But now you can, with satellite imagery, that's now low cost, you can do precision
agriculture
all again available through your smartphone, right?
And with low cost sensors, you can do so much more.
So now we're digitizing the farmers' field, right?
And putting it into action, absolutely along the way.
So what's interesting to me about this career is that I think, and tell me if you
agree
with this perspective, my perspective from wandering around the developing world is
that
the barrier to applying these is as much in people's mindsets as it is in the
technology.
The technology is more way ahead for humankind to catch up with it, requires a
light to
say, self-servingly, people who are packaging it in a way that they can understand
it.
100%.
I mean, I think the big thing, my sense, is that the cost to innovate with these
technologies
has dropped exponentially, right?
And the tools available to use the technologies has become widely available.
So now I think it's a matter of imagination.
It really is.
It really is.
It really is.
And if I can connect to what we spoke about earlier, it's about building the
ecosystem to
allow this imagination to click in.
Exactly.
The tool is there, the science is there, the ability to compute power is very cheap
compared
to before.
If somebody can sit there and say, do an exercise of what ifs, which I highly
encourage everybody
to do all the time, it's like, what if we did something crazy?
And then think, OK, to make that happen, I need to do ABC, like the examples we
were
talking about earlier.
This is very cool.
Yeah.
Yeah.
And just to sort of keep giving a few more examples of this, right?
This is impacting a whole bunch of industries.
So in the media industry, we see the same thing, right?
So what the folks are seeing right now is what happened, what used to happen in the
past.
So there was this great, there's a movie now to about the American Hudson River.
Remember the US Airways plane that lost his power and then, I forget they were the
captain,
he sort of landed it.
Right.
Yeah.
Very famous.
Sully.
Sully.
Sully.
Sully.
Sully.
Sully.
Sully.
Lances.
There's a movie about it.
Tom Hanks played Sully.
You know, a hero.
That's up.
Yeah.
This was like broadcast live on CNN, right?
Because the CNN studio was right there, right?
And it was made all the news, right?
Yeah.
Traditional sign.
Then a few years later, there was an amazing example of this treat.
So in San Francisco, you know, a plane coming from Korea, crash landed on the other
run
right in San Francisco.
And David Unwe, it was a Samsung executive who got the chance to meet, writes the
following
tweet, right?
I just crash landed in San Francisco, tail ripped off.
Most everyone seems fine, I'm okay, surreal, and he's post a picture.
So David is my classmate?
Is he really?
You're kidding me.
You know that.
I didn't know about this tweet.
It's so funny.
David was at Harvard Law School.
No way.
The same year that...
You were an interesting...
Yes.
And we met because I did a lot of work on Samsung next when he worked at Samsung.
Okay.
One.
Small world here in San Francisco.
So again, before you need a CNN, big cameras...
Yeah.
Now it's like...
It's like...
The person who's experiencing the news is part of the news, is the news story
themselves
is breaking the news, right?
And that's what Twitter did.
That's what Facebook's done, right?
That's what TikTok's done and so forth, right?
It has basically shown that the media industry, which was concentrated amongst all
of these
TV studios and all of these newspapers, is now fully democratized for what good and
for bad, right, in terms of what that does, right?
So this is part of the digital transformation.
Again, the smartphone, and then somebody's saying I'm going to build a platform to
trade
images and to trade a little bits of information becomes a media platform as well.
So this morning, when I was driving in to do this video shoot on National Public
Radio,
there was a story.
There's been a lot of angst in America, but local newspapers dying out, the
economics
are not working, and private equity firms coming in and gutting them and so on, and
it's a mess
because you have these news deserts, news deserts, which is bad for democracy and
so on.
So the NPR guy suggested, I think it's just brilliant, is, look, just take the kids
in
the high school and train them to be citizen journalists, and they wander around,
they'll
get expertise in learning how to write, to get expertise in, to interview, to
prevent,
to present balanced perspectives, and bring them into it.
Don't think of it as a calamity that we bring our hands, think of it as an
opportunity that
we can do something with.
Exactly.
So that's a democratization of news, and it's feasible because of digital media.
Exactly.
It's great.
And then, of course, all of us experienced this now globally about what happened to
the
taxi industry, right?
Grab.
Yeah.
Ola.
Right.
D-D-D-D.
Uber.
Uber.
You know, Korean in the Middle East.
Yep.
You know, all these things happen.
And do you remember, very good, 15 years ago, how hard it was to get a taxi in
Cambridge
Massachusetts?
Yeah, yeah.
I think I'll call up to, like, book a week ahead, and they'll give me, like, a two-
hour
window while maybe the taxi drivers will show up, right?
And then they'll re-manage you anyway that you've disturbed them from drinking
their donuts
and drinking their coffee and donuts, right?
And maybe they'll show up.
Do you remember that?
In Newton, we had this comogenly guy who would shout at you, if you call them.
Yeah.
You'd call somebody.
All right.
All right.
And so we were in this world, but the taxi drivers had to buy the cabs, right?
And medallions.
And medallions.
And medallions were super expensive and so forth.
And then we went to this world that Uber started, which was, like, ride-sharing,
right?
Like, you can basically have a star source with limousines, but then any driver
with lift
actually starting that, that revolution, any driver could take anybody somewhere
else.
And just think about how much the world has changed, right?
Like, you're not getting into a car of a stranger, and they can take you from a
point
to a point B, and you're not going to give them any money.
It's all in the back end.
It's all digitally enabled.
It's all digitally enabled.
And so we cut a car through your payment system.
And so we went from this world of asset heavy to asset light, right?
And we democratized who could be a taxi driver.
A democratized transportation, though, wasn't it someone awfully of cab drivers and
limousine
drivers.
Anybody could potentially participate.
You know, when I was talking to Anthony 10, the co-founder and CEO of Grab in
Southeast
Asia, he goes like, you know, they are enabling 10 million Southeast Asians to have
income,
right?
Because they now can use their mopads, their bikes or their cars to do delivery, to
do
transportation, and so on and so forth.
And so again, that's the digital transformation of the auto industry.
And look more generally, we've been sort of living through these waves over time of
all these companies coming in and transforming from going from an era of mainframes
and computers
to PC and the web to mobile to networks and AI.
And now we're in this world of AI.
So an interesting parallel that's going through my mind because we started with
thermostats
and utility companies and nests and so on.
And in some sense, what we're also living through because of digitization in the
energy
industry is sort of what the computing industry went through from mainframes to PCs
to handheld
supercomputers is that the generation and dissemination of electric power is being
democratized.
Yes.
And so you have solar panels, yes, you have small wind generating that's right and
a whole
geothermal.
Geothermal.
So everybody is going to be an energy producer and a consumer connected to the grid
digitally.
And then you have this amazing market making going on.
It's very exciting.
Yeah.
If there was, since you went to school in the last century, do you know what this,
uh,
what this image is? Well, what is it that you reminded me of? It looks like an
image somebody
with my limited skills would produce today. With your, with your iPhone, yes. Yeah,
because,
you know, I've got a little bit of old words. That's my check. It's a bad image of
somebody's
cute baby. Yes, yes. So in fact, the, this is the first ever digital image. I want
to say that
that's before I was born. That's before you were born, 1957. This is baby Kirsch.
So he's older than
you through. Not that much older. And, uh, and this was the first digital image
and, uh,
it was 176 by 176 pixels, right? So this, this started the Instagram generation,
the TikTok generation in many ways. Long time ago. Okay. Um, and the, why is this
important?
Because this sort of builds on our conversation about digital images. Once you have
a digital image,
right? You can take it and then you can infinitely reproduce it, right? You can
make it available to
anybody. And you know, this revolution killed Kodak, right? Uh, and Polaroid,
right? Because
they didn't understand that it was long about film and log film, but about digital
images, right?
And so this is, this is the digital world we're living in. But now let's take the
digital world,
these digital images and apply some algorithms, right? So I'm showing you, uh, like
a view of a
baby Kirsch, right? Uh, this old image. And then let's see, you know, what would
happen if you'd
added a neural network to it instead of said, like, what would the neural network
do, right?
Well, so then you take this image, right? And you make it look better, right? I can
improve
upon it because it has some rules about improve on it. But you can also do fun
things like age it,
right? So then you have like ancient baby Kirsch as well given. You can do all
these
things with algorithms and AI. And so what we're going to try to understand now is
that the
digitization revolution that we talked about is now big, supercharged now with
algorithms and
artists and intelligence, which is creating a whole new set of opportunities for
all of us as
well. Great. When most people think about AI, the immediate reflection and points
of connections
are things that the scientists call strong AI, which is from science fiction.
Autonomous robots,
C2PO, R2D2 doing their, my favorite, right? Star Wars, Star Trek, all that kind of
stuff.
And so, and so, so that's known as strong AI, but that's still the realm of science
fiction, right?
Where all the action is actually is what's known as weak AI, which is what you were
talking about,
right? Any activity that humans once did that computers could not do. So like make
some inference,
like, oh, is this a good image or a bad image? Is it a cat or a dog? You know,
that's known as weak AI.
And what's so interesting is that weak AI is basically a combination of statistics
that's been
around for a long time, applied to lots of data. So we talked about the fact that
there's a data
explosion happening with digitization, right? And, and lots of compute, lots of
computers and cheap
and cheap computing power, right? Becoming available. So, so weak AI has basically
changed
the world. And so it's statistics? Yes, enabled by large data sets and cheap
compute power.
Exactly, exactly. In fact, one of the first cool examples of AI was, do you
remember a time when
you had email and there's like lots of spam coming in, like lots of spam was coming
in
and then overnight it went away. The reason why it went away was because some
people got smart and
said, I can just use a logistic regression, right? If you take statistics, you
know, logistic
regression, right? And differentiate between good email and bad email, one and
zero, one and zero,
good email, bad email, I can have the rune label, a bunch of bad emails from me
saying, this is
spam, this is spam, right? And know the remaining is good email. And then that
gives me the
characteristics to then be able to predict if a new email is good or bad, spam or
not. And that's
all weak AI being put into place. And by the way, right, all of us now use weak AI.
So if you're
using a smartphone, right, if you're watching TikTok videos, if you're using WeChat
or using
WhatsApp, right, if you're on Facebook, right, any kind of online platform using
Netflix,
using any kind of online platform, right, weak AI is powering all those things.
Okay, when you use
Google, when you use Microsoft, it's all DD, you name it, it's all using weak AI
today. So we've
become consumers of it, we just didn't know that was going on. And what you see
through is a ton of
hype, right, a ton of hype about AI. Now, what do you think there is that like,
like, you know,
you're a quasi economist, what, what do mathematicians are economists? If there's a
lot of, a lot of
activity, what do you think might be going on? People are imagining and projecting
their hopes
and aspirations and worst fears onto some imagined future. Yeah, exactly, right?
So,
so one thing that's happening is that the demand for AI is increasing, right?
Because all of a
sudden we're imagining, oh, you know, could AI make us better teachers, you know,
could AI make
better dentists, could AI make better drivers, right? So we're imagining new
opportunities for AI,
but or new fears, or new fears exactly the same. Would I be replaced as a
Oh, huge. HBS professor. Yes. By some AI bot that simulates a class discussion.
That's right. But it might be good for us in our aspire mission. Yeah. If we had
the
rune bot that could be a million people, you'd have to like, absolutely. We know
won't bring
me in every weekend. That's right. Do a session. We could be bots. You don't know.
So, so, so, so certainly there is, you know, like this, this view of the demand for
AI
increased, but also importantly, right, the supply of AI has also increased in the
sense of
the cost. The cost to innovate with AI has dropped dramatically. What would take a
team of a thousand
people doing something and I'll be done with 10 people. Yeah. Right. And over the
last 20 years,
we've sort of seen this happen where both again, compute has been, computer power
has been
democratized access to the algorithms through open source has happened as well.
And lots of data are emerging as well. And that's why we see so much, so much hype.
And so, you know,
you know, I claim that we're sort of entering this age of AI where much of our
personal lives
is being run through AI systems. We just don't necessarily know. Yeah. It's not
explicit. But
with the introduction of chat GPT, most recently, you know, it's now become first
center of what we
could do with it. For me, one of the most insightful visions of AI, you know, came
from this photo.
So what is this? What is this photo? This is a Dutch photo Rembrand. This is a
Rembrand. Yeah.
It's a Rembrand. In fact, this whole image was generated by an AI system. Okay. And
when I show
this image at a conference in Australia, this, this older than you gentlemen stood
up and said,
Oh, I know what that image is. That's a Rembrand. I've never seen this Rembrand
before. That's a
Rembrand. I said, Sir, how do you know? Well, I'm the curator of the Australian
Portrait Gallery.
And I did my PhD on Rembrand. So this looks like a classic early Rembrand. I'm
like,
Sir, let's watch this video together. And you tell me if you think this is the real
Rembrand.
So we're going to watch a video about how this because we're talking about AI, how
AI have
generally used this image, this image.
One of his great achievements, one of Rembrand's great achievements was to portray
human emotions
in a much more convincing way that artists had before him and in many ways for all
time.
As I and GV 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 Rembrand
used his paints and his brushes to create something new.
The first step was to study the works of Rembrand in order to create an extensive
database.
We gathered 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 Rembrand paintings were portraits,
we analyzed the demography of the faces in these paintings, looking at factors like
gender,
age and head 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,
and our 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 Rembrand Rembrand.
We took parts of the face and we started to compare them, and then based on this,
we were able to create a typical Rembrand 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 hive map is essential to make the painting a painting.
We incorporated the hive map into the painting and printed on a 3D printer
that uses a special paint base, UV ink.
It printed many layers, one on top of the other,
which resulted in the height and texture of the final painting.
Sometimes a magical moment to see a painting for the first time,
even if it's completely generated, for me it is something special.
I would have believed, if I would have saw it in a museum, that it would have been
a real Rembrand.
Just what I haven't seen before.
So Therun, what do you think?
Pretty compelling.
Amazing, right?
It makes you think about what's, you know, soon you'll have like 3D
holograms, computed by AI, projected, and we won't need to be here.
You know, also what's interesting is like, you're on the board of directors of the
MFA,
you know, the Museum of Fine Arts.
I stepped down, but I was.
You were, you were on the board of directors for the Museum of Fine Arts.
And now art and art history is going to be mediated through AI, right?
It's incredible, right?
And when this portrait got released, there were two types of examples, reactions.
One reaction was the best thing ever.
Oh my god, how incredible, how great, right?
And you can imagine also what the other reaction was, right?
The worst thing for arts and humanities and creative expression and that
everything's been
automated, blah, blah, blah, blah, blah, blah, blah.
Right, right?
And so this quote by one of the leading art critics in the garden said,
you know, the digital Rembrandt, a new way to mock art made by fools, right?
And the point to note is that as these technologies become more and more
widespread, right,
we're going to have this dual reaction, right?
A set of people are going to love it, embrace it.
And I said, people are going to resist it and push against it.
And even for us, you remember how many of the folks out of the Harvard Business
School
were against online teaching?
Yeah, absolutely.
Right?
No way we can do the case method online on Zoom.
Oh my god, how terrible.
And then COVID happened and then, oh my god, we can teach us it.
That's the only way we could do it, right?
And so this, what I want to tell people is that as cool as this technology is,
and as cool as the early adopters will take it, there will be resistance.
Yeah, so, you know, I just want to pause here for a second because
this is a, I think, a really important point.
And we've already mentioned it before, but technology is not per se good or bad.
Yeah.
It's just is.
It's like a hammer.
If I take a hammer and I hit a nail, it's productive, hopefully.
If I take it and I hit you on the head, it's not productive.
So it's a little bit like that.
And reflecting back on, so I'm huge museum lover to my, to my kids, horror, because
that's how you do the drive-down museum.
I know.
It seems they hate it.
No, Papa, not one more museum.
So, but I've seen this in museums all over, including at the our local MFA,
which I love as an institution, is that there is such visceral fear.
So, not necessarily coming from, do I have a job and not have a job?
Or what do people, you know, think about my skill sets?
But just coming from being relatively uninformed about what's going on.
Yeah.
And I think there's a deep human reality there is that when something completely
new happens,
relatively few people take the time to find out enough to then realize whether
they should be happy or not.
Exactly.
And to me, that's, you know, we're trying to educate leaders.
We're trying to educate change agents.
Yeah.
And we're aspiring with aspire to do that for millions of young people around the
world.
Yes.
Is that you have to take the time to lean into this newness before you form an
opinion.
Yeah.
Yeah.
But also importantly, I think, I think what you're saying is that for our
participants,
yeah, this is going to be the reality.
Like we grew up in a world where calculators were banned from classrooms.
Yes.
Right.
And then the internet happened and all that kind of stuff happened.
And so we've been through these waves.
Yeah.
But all of them are born into this.
And this is going to be like the oxygen.
I started teaching at Harvard College two days ago in my course on economic
development.
And the first thing I said was, please use Chachibidi freely in whatever you want.
Just let us know how you're using it so that we can adapt and do it together.
So I think that I'd like to model that kind of behavior.
Exactly.
And for our participants embracing this, understanding it, and then making a
judgment.
And by the way, this is not just what the technology is.
This is for everybody.
Everybody needs to understand it and be able to put this into place.
Because whether you be in the arts, humanities, health care, law, politics, this is
the core,
this is going to be the core part of things that you need to understand.
Let's take a step back. I love this book by Azim Azar called The Exponential Age,
and this is the
key point for us because what's happening with Azim claims is that there's a set of
technologies,
AI, synthetic biology, solar, which you know really well, and 3D printing that have
exponential
capabilities, they grow exponentially. And with them are companies and
organizations that grow
exponentially, and often consumers that grow exponentially. And then there's a rest
of us that
grow linearly, okay, our organizations, our institutions, our societies grow
linearly. We actually don't
have a hard time thinking in an exponential, exponential ways. We see that in the
institutions
that we're embedded in, you know, this is a what, 400-year-old institution, and
unsurprisingly it
has trouble thinking exponentially, right? There's a linear growth is like, you
know, two or three
percent a year, five percent a year, like our undergraduate class hasn't grown much
at all in
the last hundred years, you know, relatively speaking compared to how big the
population of
the world is, or even if America is. And so, and so what this does though is your
your mathematician,
what happens when you have a system that's grown exponentially versus a system that
grows linearly?
Big disparity. Big disparity happens, right? And so, so this is the world again,
that we need to
be thinking about that are all of our worlds are going to have this kind of digital
AI divide,
exponential organizations, exponential systems, exponential expectations, and then
linear systems,
linear growth, right? And there is in this divide, this this this gap, right? There
is
opportunity, but also risk as we move forward. And so, so for me, you know, as I
thought about AI,
and as I was working with my good friend, Marco in CD, my mentor as well. I don't
mind too.
You're as well. We'll talk about that. We'll talk about Marco. We love you Marco,
sometimes.
Yeah, I love you more now because you've been away from me for 20 years. Okay,
we'll come back. He's like, you know, he's a faculty member with us. Yes, you can
see we love
using each other and everybody else as well. So tell us about as we conceived with
your your
leadership, we conceived a spirant shoot as a digital first organization. Yeah,
yeah, tell us
about that. Yeah, so so for us, the book I wrote with Marco in 2020 really sort of
was based on our
decades worth of research that we did together saying like a new type of company
was emerging.
And this use of a company was applying digital platform thinking and AI
systematically across
its business model, how a company creates value, how it captures value, but also
importantly,
it's operating model, how it delivers value, right, how it serves many customers,
we call that scale,
how it offers them many different products, we call that scope, and then how it
keeps improving
and learning. So in many ways, this is a foundation of a business, you have a
business, you have a
strategy, a business model, how you create value, how you capture value, and then
how you do actually
deliver value to the operating system, right, through scale, scope, and learning.
And the example
I'm going to use is what we covered earlier is an offshoot of Alibaba, which is the
edge group,
right? That's fabulous, right? So, edge group is an amazing example of a company
that became
AI first. So first of all, it's a, you know, as, as they were going to explain
earlier, Alibaba
basically solved the trust problem in e-commerce through escrow. Escrow comes
along, and that comes
through their app called Alipay. Alipay basically starts to do, to do transactions
on Alipay platform.
They spin off Alipay into a separate company called Ant Financial. And the reason
they call it Ant
is they go like lots of small transactions, but Q and together for the for the
common for the common
guy. And what you see is this incredible exponential growth, right? They go from
20, 30, 40 million
people to 1.2 billion users, okay? That's incredible. It's short amount. It's
mostly in China,
China and South East Asia, 900 million in China, and then the rest in Southeast
Asia and so forth.
And what's exciting about this company is the following and scary about this
company is the
following, right? 1.2 billion users, 15,000 employees. Almost none. The scale,
right? The scale
that you can serve, 1.2 billion people is mind blowing. Do you know like what
percentage of the
15,000 are engineer types? 50%. 50%. Half the company are. So that's really
interesting too. Half the
people are technical people, right? But but just think like whatever bank you have
here in the US,
if they were serving 1.2 billion users, how many employees would they need?
I don't know. Several million, right? Several million. It'd be bigger than the
Chinese army.
You know, because so the point I'm making here is that in order for you to have
this kind of
leverage of 1.2 billion users and only 15,000 employees, you have to have digital
and AI at
its core, right? That has to be at its core. And that's what Ali and Kripa shown.
So they went
crazy. They went, they had and financial, they built UBOW, they built their own
bank, they built
a cloud system as they were growing the company. And what we see them do, what
becomes so interesting
about them is that their whole strategy was, we want to be a fin-life provider. We
don't want
to be a bank. But whenever somebody spends money, we want to be the one in the
middle. We want to be
the one that does the transaction, participates in the transaction. Because what we
care about is
the data. Just like Nest, we care about the data that's happening and that's being
pursued.
Can you say a word to our participants about this idea that hasn't come up, but I
think
it's central to this exponential rise of increasing returns. What does it mean to
have
increasing returns? Yeah. So basically when you have a system like Antsgroup, which
is
fin-life, right? So what you're seeing on the screen is the app is not just a
banking app,
but it connects to a whole range of services. And in this whole range of services,
what happens is that as the number of users increases, the number of other
providers that
want to be on the platform also increases. And that in turn, again, feeds back. So
let's go back
to a more simpler example. If you think about a platform like Uber, Ola or DD, Uber
or Ola
and DD is useless with just one driver or one user. It's not going to create a lot
of value.
And even five drivers and five users are just not good. You need lots of users and
you need lots
of drivers. But as the number of users increases, the number of drivers will also
increase.
So the value of DD or Uber will increase as a number of users and drivers
increases.
But the value not only increases to Uber, but to you, right? Because now there's
more drivers.
Yeah, I can find a car exactly. And similarly, for the driver to, I can find
passengers to,
right? So that's the exponential network effect story. Also the incremental data
generated by
the next driver joining or the next user joining. It's so valuable. Exactly.
Because there's already
data. Exactly. And that's why you sort of see why they went ahead with this model
of like,
let's make sure we connect to as many other providers of financial services and
wherever
money is being spent, because the increment value of the data for this is going to
increase.
Yeah. Right. And that translates into what we sort of see happen, right? So for a
company like,
and they have this traditional viewer product value, like, oh, how nice is the
product, how well
does it work, right? The product, the form factor and so forth. But there's also a
narrow
value. If it's sitting on an ecosystem, the value will also increase, right, often
exponentially.
And then because you're now, you have now got data on every single transaction,
every single movement, you also have data value as well. And that all those
basically sharpen
the value curve that these companies generate. And that's why they're growing
exponentially.
All right. And then similarly, what we see happening is that there's an operating
model,
because that's not as far as you were doing this, but we got to build this damn
thing, right?
And so the core of this is, is a system, which is all about the data. So you have
this ecosystem
around you. And the logic is the more data you have, right, the better algorithms
you build,
the better algorithms you build, the better services you offer, the better services
you offer,
the more users you have, the more users you have, the more data you have.
And so it's a virtual cycle.
It's a virtual cycle. And I think many of us in our consumer lives face that
already, right, like
when you, when you use a Google, when you use a Spotify, when you use a music app,
when you use,
right, it's this, you're benefiting from it, because they're continuously getting
the data,
building the algorithms that come along with new services, right? So that's the
logic underneath
their organizations. And the core thing behind these companies is what we call an
AI factory.
The AI factory is a really important concept. And it's not just for large
companies, it's for
any enterprise, right? The notion is that we have a systematic solution to data. As
more data
comes in, we need to process it. And we need to do three things at the same time.
We need to make
predictions. We need to do pattern recognition, right? And we need to do process
automation.
So let me pause you for a second. The verbiage of a factory, even though you said
it applies to
everybody, conjures up large scales. And when I see some of these terms, I worry
that we have to
have a lot of people who do algorithm development. But really, what I'm hearing you
say is that
you're pointing to the functionality that even one of our individual participants
should be
thinking about where, what's the data pipeline in which I want to embed myself,
right? What's the
way in which I'm going to either myself develop or induce others to develop
algorithms to help me
do what I want to do? And what infrastructure can I leverage around me, even as a
solitary
change agent? 100%. So it applies to everybody. And also these three elements of
predictions,
pattern recognition, and process automation, they're generic. Because if you think
about it,
we do that. That's how humans work. We reason by analogy, we find patterns in our
brain,
we don't know exactly how it works biochemically. But that's what we do all the
time.
Exactly, right? And so we talked about their institutional board. So one of my
former students
set up a dental AI company, because what he discovered was that in fact dentists
were really
bad at diagnosing x-rays, right? And so he said, well, I'm going to get a bunch of
images,
then I'm going to train them and say, what's a good image? What's a cavity? What's
a
not a cavity? He got the algorithms from open source, right? Went to his local
dentist and said,
I can probably help you make better diagnosis himself, right? And then basically
built a company,
right? That today, four years later, is about 40 people, but they're going to serve
a million
patients. Fabulous. Right? And so it doesn't require that much. And the thing is,
Florian has some technical training, but all the stuff was off the shelf. And this
is available
to everybody, right? This is available to us. Again, the aspiration to do
something,
the imagination, and then just to have empathy with people and trust building to
bring it all in.
But what you're doing is you're applying AI. Yeah, I found you to do three things.
Again,
predictions is about three P's of AI, right? Predictions, pattern recognition,
right? And
process automation. Just think about any kind of activity you do. There's a
prediction element.
Is something going to happen or not? Right? There's going to be pattern
recognition. What's
good? What's bad? Right? And then there's like some process that you need to
automate,
right? And so that's what AI factories do. And that becomes the core part of your
operating model.
And so let's actually look at an AI factory, right? And watch this video. Again,
this is like a
fun science fiction video. But actually, this is like real today. Like you can walk
into many
stores and experience the AI factory. 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.
Our technology will
update your virtual cart automatically. So how does it work? We used computer
vision,
deep learning algorithms at sensor fusion, much like you'd find in self-driving
cars. We call it
just walkout technology. Once you've got everything you want, you can just go.
When you leave, our just walkout technology adds up your virtual cart and charges
your Amazon account.
You'll receive a sense straight to the app, and you can keep going. Amazon Go.
No lines, no checkout, no, seriously.
I've walked into stores like this. What is it? Amazon Go, something, a store where
you walk in
and you pick up a sandwich and you walk out and you build automatically. I'm
actually remembering
that even 10 years ago, I walked into a store in Dusseldorf in Germany, run by
Metro.
Metro AG is a big, like a cashier. And they had a store of the future somewhere,
and I remember walking in and the version of this that was there at the time, and
it gives you a
sense of how technology has changed in 10 years, was pretty far out. I mean, they
basically are
a little robot, like which one is R2D2, which one is C3PO. A little R2D2 follows
you the minute you
step in, and is basically taking images of you as you walk, and is then trying to
predict what
you do, not perfectly, but they were clearly trying to get ahead of the car.
And now it's like automatic, right?
No, what's interesting, we're talking about ethics, right?
Let's just imagine how many people in India work in the retail sector.
Oh my god, tens of millions.
So you can imagine why there would be resistance in emerging markets around these
technologies,
because this might cause vast unemployment, right? And so we have this double-edged
sword,
as you said, right? Like on the one hand, amazing stuff, right? On the other hand,
it could cause vast unemployment. And so that's the kind of thing that we'll have
to keep
thinking about as we talk about these technologies, and as we deport them.
In many ways, my feeling is that they're inevitable. It's just a matter of timing,
but then we need, as leaders, to think about the negative effects, right? And
encourage our
societies, our governments to have the proper retraining programs and dislocation
programs
to help people shift over. Which, again, unfortunately,
is not going to happen in the emerging markets, because the states aren't
necessarily equipped to do
that. But again, that's an opportunity to do something about softening the landing.
That's how societies move forward, is people solve problems.
Exactly. So let's put it all together. So how does this now connect with digital AI
first business with a traditional company? So as we talked about, digital
businesses can grow
exponentially, right? The ant financial example that we talked about, you know, the
Google example,
and so on and so forth, right? And they are basically colliding with traditional
product-based
businesses that basically grow in this sort of this concave fashion, right? They
have tremendous
value initially, and then they peter out because of complexity, and this creates
the collision
of these types of businesses. These businesses are organized in very different
ways. One is
organized as a platform, one is organized as a multi-division unit, right? With
silos, all of the
price, all on the business. Even we had HBS, sort of C silos, even in our small
school along the way
as well. And so, but this has some interesting consequences, because what we see is
that you can
compare Airbnb to a Marriott, you compare Waymo and Tesla to a Ford, you compare
Ant Group to HSBC,
and even Moderna to Pfizer, right, as examples of how the world is changing. Now,
one thing to do to
look at, as you look at this graph, right, I'm going to take away the logos, right?
And just
compare these two graphs. What does this graph also remind you of? What have we all
experienced in
the last three years? The pandemic. The pandemic. Yeah. And COVID-19 was what?
It was an exponential rising disease. And the hospital capacity was fixed.
Traditional.
Exactly. So the whole non-evolving. Exactly. It's the exponential process and a
linear process,
what we talked about from a disease point of view, right? So the whole world, all
of us
experienced what happens when an exponential system collides with a linear system,
with a fixed system,
right? Lots of disasters. Guess what? This is what's happening in the business
world, right? You have
companies, and we hope that many of you generate companies, right, that can grow
exponentially,
our desire to aspire is to be the first digital AI-first nonprofit that can grow
exponentially,
to have exponential impact, right? And then we're going to hit against, right,
people that don't
know how to think about exponential systems. We already are. Yeah, we already are.
People are
like, how could it be that you could serve 100,000 people? Yeah, right. And because
we have thought
about a leverage model and a scale model as digital first. Now, we're still
learning how to do this,
put this together. But the point that I want to make is that this kind of thinking
is native
to you guys, all of you guys should be thinking about younger people. How do I
actually build an
exponential organization? How do I get leverage? How do I have lots of impact
causes problems for
others? But you're also solving problems for many more people as well, along the
way.
And it's sweet. Yeah. And so I want to quickly give two more examples of this. So
we touched upon
Moderna. We talked about vaccines before. So I've gotten to know this company
really well. And
what's interesting about Moderna is when at the start of the pandemic, they were
around for 10
years. They were just pioneering the mRNA technology. They had no drugs in the
marketplace. But they
were an AI first company. Sefan Ben-Salena alumni from our school basically had
made a bet that he was
going to make a digital first AI first company. And so when the pandemic in pharma
in pharma,
so not in a not in a fully digital world like like a Google or Facebook or Amazon,
but in pharma,
the hardest place is to do this. And what he bet on was that he could build an AI
drug design
studio with mRNA. And so when the pandemic hit, right, they took the sequence,
genetic sequence from
Wuhan, they never touched the virus is all digital, right? And then in four hours,
they created the
vaccine candidate, right? Four hours. Amazing. All digitally, right? Sent it to the
digital factory,
got it made, and then basically tried it out. It worked. He told a scientist like,
maybe you should take four weeks to just verify this actually works, right? Because
I mean,
who in pharma does something in four hours, right? Even Iran lost the companies.
And then the rest
is history, right? And so Moderna is a great example of a company that has figured
out how to-
That was set up to respond to this. Exactly. That's a digital first AI first
company.
And made this happen.
So, Karim, we spoke a lot about AI factories and exponential change and so on, and
lately,
we were either excited or shell-shocked by chat GPT, which popped up everywhere.
And the associated phrase that's become part of everybody's lexicon generate a way
to tell
us how all this fits together.
Totally. I mean, I think, I think for me, uh, generative AI is so transformational
and it's going to impact everybody. Uh, and, you know, because of the fact that
everybody's
now connected through the internet, the adoption of Janet and it's use cases are
just exploding,
and so we'll talk about that. So, of course, let's put this into context, right?
Like, how does generative AI fit into the sort of all of AI and all that kind of
stuff? So,
you know, like, what we talked to you about earlier was this notion that
sort of there, there've been these constant investments in technology over the last
50 years.
Well, similarly, as technology investments were being done in compute and computers
and PCs
and that kind of stuff, there's been a similar path in computer science to sort of
create these
artificial intelligence systems. So, there was an era in the 50s and 70s known as
cybernetics.
Uh, uh, Norbert Wiener out of MIT was the guy I had to read Wiener and Marvin
Minsky.
Did you? Yeah, exactly. Right. There were all about these cybernetics systems.
And then, then there was like a, well, it was known as that AI winter because the
cybernetics
systems went only this far, right? Right. And then you probably remember you're old
enough
through winter remember, export systems, right? I, I wrote a book called
Foundations of Neural
Networks on the first books and neural networks. Really? In 1989? Yeah. Have you
eaten it by hand
or something? I did it on a typewriter. On a typewriter. And the hand drew all the
neural networks. So then these export systems were being devolved and they went a
certain distance
and then there was under winter. And then what happened in the 2000s was basically
the internet
came about. Many of the companies we talked about, like the Google's, the
Facebook's, the
the Amazon's, the Microsoft's, the DDS, the, and so forth, the Alibaba's, the
world,
they were awash in data, right? They had digitized it. They were awash in data.
They were like,
what are we going to do with all this data? So simultaneously, there was work being
done
on neural networks, on convolution neural networks and so forth at various
different universities.
And there's some breakthroughs on how you could take those neural networks with
actually
PlayStation processors, Sony PlayStation processors and do image analysis. And that
basically
started a wave of machine learning systems to come through. So all the systems that
I talked
about earlier that we take for granted, like our map systems, our music
recommendations,
our book recommendations, our movie recommendations are powered by our spam. Those
are all powered
by these good learners and all the speech language processing, all that kind of
stuff is all being
powered by good learners. And now in several winter, which many people predicting
was going to happen,
we're now in this generative AI systems. And the way the generative AI systems work
is that
they basically generate text. They take an existing set of data and instead of
predicting
the next action or the next sales thing, they go, what should be the next word be
or the next
pixel based on past patterns, past patterns, exactly. And they've got tons and tons
of these of
these patterns. And so in 2022, there was this major breakthrough. And again, this
is because of
these things, data, compute, right, and statistics, right? The same fundamental
things tree,
now it's just being supercharged, even more and more and more and more. And the the
this
exponential stories actually like really becomes apparent when we sort of think
about what's happened
majority of AI systems. So standard machine learning models, right, starting with a
basic
regression form, right, up to deep learning had about 10,000 parameters, you could
run a regression
system like 10,000 variables, right, did some things deep learning systems, right
showed up,
right? And that gave us some breakthroughs and image analysis and so on and so
forth,
could have up to 10 million parameters. The first version of Chad GPT 3.5 had 175
billion parameters.
Oh my God, that's right. So many orders of magnitude. Exactly. And so this is what
happened.
You basically went from algorithms doing one thing and one thing alone to not
create these
pre-trained models that were multi-mole, they could do many different things at the
same time.
And that's the big breakthrough. And it's again, the same exponential curves that
we sort of talked
about earlier, the same thing is going on. And that's again, a key thing for you to
realize that
these things are going exponentially, their use cases are going exponentially, it's
a huge opportunity.
And people who are not leaning into it are. Yeah, again, we left behind.
Are in real danger, we left behind. And the point is you don't have to be a data
scientist.
No. You don't have to be an AI scientist to learn this. Look at us. But the point
is,
you need to understand the technology and then do the translation to your context.
To your context. To your context. Very important.
And so let's start and talk about it. So earlier we showed the next Rembrandt.
So I'm going to ask you a question. Do you watch Bollywood movies?
Every day. Every day. Three hours a day. Oh my God. Bad movies.
So the time. And all the singing and essay. Can you sing it? I want to ask you.
So in your household, what's your favorite Bollywood actor?
Well, my favorite Bollywood actor is some really boring art scene.
I don't know what you're going to bring that. That'd be like, who are the masses?
My wife's favorite actor, Shah Rukh Khan. Shah Rukh Khan. Okay.
Now imagine the question. Imagine I said, okay, let's create a Rembrandt style
image of Shah Rukh Khan.
It doesn't exist. Shah Rukh Khan is not around when Rembrandt was around, right?
And it would be hard for us to get him to go back in time and time travel has not
been invented.
But we could potentially ask a Genevieve AI system to give us an image of Shah Rukh
Khan as imagined
by Rembrandt. Let's do it. Ready? Ready? Okay.
Ready.
When you watch a lot of Shah Rukh Khan movies, what do you think?
That's pretty amazing, especially the one on the upper right and the bottom left.
How long did it take for us to do this? This is the classical Rembrandt portraits.
Classical Rembrandt portraits, right? 30 seconds. Not even on a journey.
I just put in a prompt that showed me Shah Rukh Khan, Bollywood actor, and the
classical Rembrandt
post. That's it. 30 seconds later, we got it.
So what it's doing is it's trawling and looking at past images of Shah Rukh Khan,
thousands, millions, tens of millions of that.
It knows what a Rembrandt image looks like and then creates a synthetic image.
This is a fake image. It does not exist.
Right. Yeah.
It doesn't exist now. It's not exist now, exactly.
But Shah Rukh Khan never sat for those portraits, right?
It's unbelievable. So let's do one more, right? I also said,
you know, I'm spending a bunch of time in New York these days. There's lots of
hipsters.
I said, what would Shah Rukh Khan look like as a hipster? Ready? Ready?
Let's do it. Whoa.
That is even better. That is even better. All four of those are,
you know, I would believe that they're the real image.
Yeah, right.
That somebody sat in.
Yeah. This is the Shah Rukh Khan, as imagine, as a New York hipster with the
tattoos,
with the beard action going on, with the fancy hair.
Right. Sure.
And I think maybe you should send it to her.
Because that homage is a homage to the tattoo Shah Rukh Khan.
Again, the point is the cost to innovate has dropped. Our imaginations can go wild.
And that's the jereve-ai moment, right? Like we can now imagine things,
create things that have not existed before.
So let me put this in the audience.
They are all, I think it's important for people to, and correct me wrong,
that they're all extrapolations of pre-existing patterns that are embedded in past
experience.
Exactly.
Is that extrapolation?
That extrapolation is generative.
There's combinatoric explosion.
There's no way that all of us have seen all possible combinations.
Exactly.
In that sense, there's a new generation of stuff going on.
Exactly. And it's simply happening in text,
simply happening in video,
something happening in images, and you name it.
And by the way, this approach of me putting in a prompt saying,
I want this to create this,
I can do the same thing with like a human protein.
I can say, show me a protein that looks like this with these kind of things.
And the jereve-ai systems can make a protein that has never existed before.
So that's the power of these systems.
Like we can now start to do these at scale, right?
Which is quite unbelievable.
Let me put this into context for us and take you back to the last century.
Okay.
Do you remember the symbol that I'm showing you right now?
I don't know.
Yeah, this was the symbol of the first web browser.
The Mosaic.
Ah, okay with the Mosaic, yes.
Yeah, Mosaic.
So in 1992, last century, the first web browser gets invented.
Yes.
And what the web browser did was that it showed us
that what would happen when the cost of information dissemination goes to zero,
right?
And I remember during the first example of it was like the Oxford Coffee Pot,
right?
Some researchers were lazy at Oxford University.
They didn't know if the coffee was full or not in the coffee pot.
So they put a camera on the coffee pot and then they put it on the web.
So they could see what it was empty?
Yeah, if you're not.
There's empty, I'm not going to cross them.
Why would I go get a coffee, right?
Somebody else would do the work, right?
So I'm not going to waste an effort and trip, right, to do that.
No, the thing about how I went for me when I saw that was like, oh,
the cost of information transmission has gone to zero
so that an obscure lab with an obscure coffee pot could have global distribution,
right?
And that basically triggered the last 30 years of innovation and internet.
In browsers.
In browsers, Zoom, all of that is based on this notion
that the cost of information dissemination, you know,
we're utilizing many online tools, right?
YouTube and so forth to show all these videos, right?
The cost of information, just imagine if we had to build our own transmission
studios
to broadcast our videos.
Very difficult for us to do.
And so when the cost of information dissemination goes to zero,
there's a straight line between that and things like
Baidu and Google emerging because then there's a bottleneck of quality, of search.
Perfect, right?
You understand that.
Similarly, many of you, all of you are too young now, I think this,
but in the 1990s, there's a thing called Napster.
Napster allowed us access to music, right?
It lowered the cost of music distribution to zero.
Do you have illegal music sitting in your...
Did you use Napster?
And never admit to it.
Okay, all right.
We won't talk about Napster and what Napster.
Napster was illegal, I shut down, because people were just creating music
with copyright transmission fees.
But Napster showed us what would happen when the cost of music distribution goes to
zero,
and then we have Spotify and iTunes come out of there, right?
Similarly, the iPhone, iPhone showed us what happens when the cost of software
development
and software distribution goes to zero, billions of apps get created.
Anybody could be an app developer, right?
And then we get things like TikTok come out of there, right?
And so the question is, what is Chad GPT doing?
What is Chad GPT doing?
And you know, my view is that the cost of cognition goes to zero, right?
Cost of thinking, right?
Just think about what I just did.
In order for me to have done those images myself with Shah Rukh Khan, right,
I'd have to go to art school.
I can't even draw a circle, right?
Spend time learning all those techniques.
Years of investment, right?
And then, right, and then finding Shah Rukh Khan,
imagine what he might look like as a hipster and then drawing it out.
That has all, or even the comparison of the amount of cognitive effort needed
to create the next Rembrandt by all those people, right?
I saw how that reduced already from the TU Delft example.
Exactly, to this, right?
And so this is like the claim I'm making is that the cost of cognition is going to
zero.
And so the way I think about this is, from a pure economics perspective,
what happens when the price of something drops and the quality increases, right?
Well, the supply code shifts out, right?
Yes.
If you put it at any given price, you can produce a lot more, right?
Right?
Which is amazing, right?
And so that's the world we're living in.
And my view at the room on this is that as we think about
folks that are watching our videos, the kind of work we do,
is we often think about, like, for any task or any profession,
there's a distribution of talent, there's a distribution of skills.
And what we hope for, of course, is that this is, you know,
this is often normally distributed or Gaussian distributed, right?
And then we hope that we're influencing people on the right tail of the
distribution, right?
But then what would happen, you think, if all of a sudden,
we have this AI superpower, which is lowering the cost of cognition,
what do you think will happen to a curve like this?
It should all shift to the right.
Shifts to the right, right?
Why would it shift to the right?
Because now all but everybody has access to it.
Yeah.
Everybody has access to it, right?
Right?
So about the size of your level of capability, you're now better off.
You're better off.
Exactly, exactly.
And so that's the view that I have, that basically,
what's going to happen with these tools like generative AI,
is that the entire population shifts to the right
for any kind of a task or a job, right?
And then importantly, the left tail gets censored
because you're now at the average of wherever the AI is at.
Yeah, that's right.
You don't have the left tail.
You don't have the left tail.
You have a much further, right?
Yeah, exactly, exactly.
And what we're seeing in our research, the rune,
is that there are three effects that's going to happen with this.
One is like a scary one, a replacement effect.
So we wouldn't need as many workers for any given task, potentially.
The second is augmentation story.
And augmentation, the way I think about that is like the
Marvel suit for Tony Stark.
So Tony Stark wears a suit.
He's a mortal.
He's a human being.
But now he can play with a god named Thor, right?
No superpowers, but Thor is superpowers.
It's just like we can have generative AI as our AI powered suit.
And then transformation, job, nature, themselves will change.
So like I've often thought about this like, what does this do?
So like I imagine that Norse practitioners, right,
can be as good as doctors with these tools.
So we just imagine there's a shortage of good doctors
around the world.
Now, Norse practitioners with these tools
could be as effective as doctors, right?
We talked about institutions.
Lawyers are often needed to build good institutions.
We need the rule of law and interpretation of law.
Well, lawyers are also scarce and expensive in most countries, right?
Well, paralegos could now all of a sudden have access to the work of law, right?
And so that's the exciting thing that this is actually enabling for many of us.
And so what I've done to ruin is I've actually gone deep into this.
And our research team here has thought about this in terms of like,
could we imagine this happening in a real life situation?
Like let's not do toy examples, but let's put this into real life.
And what we thought about it is like, there's some task that the AI is going to be
good at,
and some task that the AI is not going to be good at.
And what happens in a workflow environment?
So we're still just to go back to your very first life.
It's still in the world of weak AI.
We're still the world of weak AI.
We're not in the strong AI world.
Remind up to spins what was strong AI.
Strong AI is a complete autonomous AI system.
We're not anywhere there.
We're not talking about weak AI.
This is still, Jeremy, I still weak AI.
Just more powerful, that's all it is.
Just talking about statistics plus data plus cost of compute going down.
That's all it says, all the magic.
The only thing is, we have this exponential curve so that the power of the
algorithms
have increased, loss of data, loss of compute.
Perfect.
And these kind of crazy things can happen.
It looks like it's sentient, but that's a statistical illusion.
It's not.
It's basically parroting what's been done before.
But it doesn't, as you said, from a mathematical perspective,
the number of combination is infinite.
That's right.
So this is great.
Right?
And that's what's happening.
So we did the study trying to understand things that AI is good at,
and things that AI is bad at.
And we ran this as an experiment at a management consulting company
to see what would happen when you...
To the consultants.
To the consultants.
Right?
And we ran this as a randomized control trial.
So we have a control group.
We have a group that got chat GPT.
And we got a group that got chat GPT in a little bit of training.
And here's what we discovered.
So in a core task, which was like strategy development task,
this is what we got.
Right?
So the blue curve that you're seeing is the control group.
They did not have any chat GPT.
Okay.
Okay.
The x-axis is the score, their performance.
And then the right curve basically predicts what we said.
What we said.
Right?
That basically...
Everybody will.
Everybody shifts the right.
And the shift to the right is quite dramatic.
Basically the person that is of median skill,
right?
Median skill in the control group can get boosted to 95% in quality.
Okay?
Now this is an elite natural consulting company.
So everybody is basically smart.
Everybody is very smart.
And they're even more smarter.
Yeah.
Which is kind of crazy.
And so the effect is that those that are on the lower part of the distribution,
those that are not as high performing, right?
They get a 45% boost in their performance.
And those that are at the right tail of the distribution,
the race bar get a 10% boost in performance.
So even the right tail actually benefits.
So that's the good news.
So it also collapses the distribution.
Yes.
Somebody.
Yes.
Exactly.
Here's the bad news.
For a task in which chat GPT was not good,
what happened was that the control group did better.
Because the humans, right,
believed for a chat GPT gave them.
I see.
Right?
And did not apply their critical thinking skills.
So is this related to hallucinations?
No, no.
So one is like, again, what we talked about is like,
there's some things that GPT is good for.
So this was a task where it required lots of numerical analysis.
And GPT, without the right, plugins does not do that well.
So it gave them the wrong answers.
And people have to have no way to tell with this wrong.
Well, they could have done their user, they could have used,
they had the data sort of,
you're sort of seduced into thinking that it's giving you risk.
Exactly.
Exactly.
Because it's so good, you're seduced into thinking, it's like that.
So the people with GPT did worse than those without.
Interesting.
And so my postdoc, Dr. Fabrizio Delacroix,
causes a sleep at the wheel, right?
That basically you fall asleep at the wheel,
because then you think the AI is too good.
Yeah, exactly.
And so this creates a warning sign for us to say,
yeah, I've seen a lot of great things,
you can do many more things with it,
you want to adopt it.
But be careful, think about it's proper deployment.
And nowhere's good at, and nowhere's not sure.
The key for you all is to not shy away from it.
Embrace it and get good at it,
because that's the way the world is going.
And even if you cannot get good at it,
even some familiarity helps.
Well, can I give you a secret?
Yeah.
For how to get good at GPT?
And that's how you're going to say, you're not a new good at it.
No, no, well, I know that already.
No, but the secret is the following.
YouTube.
Yeah.
YouTube is full of all these incredible free tutorials.
Fabulous.
On mid-journey, on the art tools, on the video tools,
and on the prompting tools needed for general AI.
I, my big learn lesson was,
I'm spending close to 10 hours a week
on watching YouTube videos of how to prompt stuff.
That's really interesting, yeah.
Right?
And I've become good at it,
because I've just watched a bunch of tutorials
from people around the world, right?
And guess what?
YouTube does.
It sees me why some general AI videos
are more, right?
And then more.
So I've done that right after all.
So I really encourage all of you.
This is free to be available.
And you don't even need chat GPT.
You can use Bing, and we'll do everything on Bing AI.
Bing is freely available for Microsoft,
which is powered by chat GPT.
And so that's the cool thing, right?
So that's where the learning imperative doesn't go away.
Anybody in any field can adopt this,
and think about their use cases on this.
Fabulous.
So, Karim, that was a fascinating tour of, you know, and financial, Rembrandt, Shah
Rukh
Khan, chat GPT, Moderna, it's really cool.
But I also sort of worry about guardrails and, you know, we've seen this
technological
explosion many times in human history, and almost always there's a period of
intense
turbulence, and society kind of figures out what the rules are, the guardrails are,
and
eventually there's some contestation and we agree on them.
So where do we add in that journey with this?
Yeah, look, I think we're in the beginning stages of this, and, you know, the work
of
them at Marco, we've really thought a lot about, like, are there new rules emerging
for us to be thinking about it?
So we've started to sort of walk through as a way to summarize what's going on.
The one message I want to make clear for people is, you can't, this is inevitable,
you can't escape it, right?
And the best thing to do is to, I was talking to a lawyer who advises Fortune 10
companies,
so like the top end of the tours, and I said, with all this AI stuff, what are your
advice
on recognizing your CEOs, and your boards, and so forth, are you saying, like,
wait?
Or are you saying, run, run to this?
Go embrace it.
It goes, without a doubt, they all need to embrace it, they're not to step back,
but
to embrace it.
And I think this also matters for all of us, like, in our research, we're embracing
it,
right?
And how we teach, we're embracing it, as you said, you love lying in your course.
And so it's really important to go out and have that attitude, but then be
cognizant
of the broader rules.
And so the first set of things is that this is systematic change, this is systemic
change,
and three simple laws, basically Moore's law, which we've talked a bit about in
terms
of compute, right?
Metcalf's law, which is about network effects, and Barabashi is all about hub
formations,
all those laws, and with algorithms, all those laws come together with big
implications
for our society and our economy.
And the entire economy and systems are being hit at the same time, right?
And it's not just a thing where only the Western developed countries are getting
hit.
It's everywhere, because with the internet, AI is everywhere, right?
So just know that this is systemic change happening everywhere.
This is going to create turbulence, right?
We're already seeing this in terms of the rise and fall of Bitcoin, the rise and
fall of
companies, and that kind of stuff.
The retail park clubs, as we talked about.
One place where we're seeing it play out is speaking of Bollywood, is in Hollywood.
Yes.
The strike, the strike that's going on with the creative industry is people are
really worried
that do we really need actors?
Yes.
Can we not just create images and generative AI and...
Writing scripts, right?
I've written like five act plays, which are GPT.
The other thing is this universality, in the sense that this is no longer about the
tech
industry.
It's everything.
Right.
Education, right?
Retail, healthcare, finance, you name it.
And this...
So this universality, these elements of network effects, the elements of data, the
elements
of exponential growth, that's in all industries, right?
So you should just be...
And that's by the way, our thesis, like it applies to us in our aspire work, right?
This is universal.
The fourth thing which is kind of interesting is that we're seeing the economy
recombine
in interesting ways, right?
Like who would have imagined, right?
The Tencent is also one of the largest platforms for doctors, right?
So this is the...
Once you are in digital signal land, you can start to do many more interesting
recombinations
because in the end, everything is data and digital, right?
And then you can apply AI to it, right?
And so this recombination where we thought these industries were separated are in
fact
getting all the diversity of network.
So my big hypothesis for education is like, you know, we talk a lot about us, these
amazing
elite universities, we're so lucky to be as Harvard professors.
But you know, like the action for learning is on LinkedIn and on YouTube.
So we're cute, we're serving 12,000 people on the Harvard Business School campus.
But the real impact is on YouTube and on LinkedIn, right?
And so this recombination is happening and many of the old institutions and old
organizations
like ours, like the joke I make is like Harvard Business School is 115 years old,
but it's
inside of the oldest corporation in the Western Hemisphere, which is Harvard
University, Harvard
University was the first corporation in the Western Hemisphere, right?
So even we are going to be faced with these kind of recombination challenges,
right, on
the way.
And what we see is that the economy is getting constituted into a few firms,
whether they're
becoming dominant firms, aggregate leaders, based on your earlier process, right,
that
these aggregation systems are showing up and they're going exponentially in many
ways in
their power.
And then so what matters for us is ethics.
What matters for us is to really focus in on just as I can scale the benefits
exponentially,
I can also scale the harms exponentially as well, right?
So we have to think about bias in these algorithms, right, in the US, they don't do
well with
brown images because all the training data set has been based on white images.
They don't do good based on genders, right, mostly male.
And so there's bias built into the algorithms and so you have to combat bias.
There's these questions about transparency and control and so on and so forth.
And what I say to all my leaders emerging as well as established is that this is
not
a problem for your lawyers.
This is a problem for you in the design of your business because you have to make
this
a core thing.
I don't know if you're sort of seeing this with the executives you consult with as
well.
So I'm seeing two very different reactions and it goes back to the graphs that you
keep
putting up of exponential losses in here.
I'm seeing a lot of people who are cognitively disabled as in not leaning into it,
reassuring
themselves that this is not applicable to them because they don't see it, the
immediacy
of it and they don't appreciate how fast it's changed.
Well that's an opportunity for everybody else.
It's an opportunity for everybody else and it's going to be trouble for, I mean,
even
in this country you have traditional industries, lots of manufacturing in this
country is still
stayed, the utility industry is very old fashioned.
So you see it everywhere, so that's one.
And then the others I'm seeing those that are predisposed to be, to lean in maybe
because
they have an attitude of experimentation and optimism.
So I think it's a, there's real agency in human response or organizational response
to it.
And I'm not going to comment what I see at Harvard because we're sitting, it's a
very
mixed bag I think.
Very mixed bag.
But you're right.
But on this digital ethics issue, you emphasize it, it really, particularly in the
places
that I hang around the developing world, it really is a first order issue.
I mean imagine.
Everywhere.
But imagine the two billion people who don't even have never been to school.
How are they even going to process all this?
Well I mean, that's so interesting, right?
Because one of the lumps from our school, Salkan, has Khan Academy, which is an
amazing
impact.
But now he's putting gender of AI, the core of it, Khan Migo.
So now you can personalize the education, the tutorials.
It's not giving you answers, it's helping you think, right?
This is the cognitive tools, right?
And so I'm like super optimistic because all of a sudden, every kid, if I just give
them internet access and a device, every kid could potentially have a person that's
a tutor for any subject that they want.
That's true.
Right?
Or similarly, I can health care, what we're seeing is that what the health is doing
in
dentistry, right?
Now we can make that a diagnosis that's really accurate.
And in fact, oral health is the vector into the rest of your health.
That's right.
My wife is a dentist.
Exactly.
Oh, so many diseases manifest in the oral cavity.
Exactly.
Exactly.
And so this is the upside, but we can't forget the downside, right?
And that's up to us as leaders to keep that in mind.
That's great.
And then by the way, you know, this problem is haunting both the AI first companies
because
they're not paying exactly.
So they're not paying attention to the ethics and also to the traditional firms as
well.
Yeah.
Right?
And so we're seeing this over and over again, that ethics matters.
You have to make it a first principle way to work towards solving these problems.
And so then finally, you know, we are in this era of vulnerability, right?
People are going to be vulnerable and us as change agents in our societies can't
forget
about the people that are going to be this proportionally impacted and either come
up
with private solutions, public private solutions, non-profit solutions on this.
But we're going to see a lot of vulnerability.
This is why, as you see as part of the HBS's D-Cube institute, which you're
running, we
have the technology for all that.
That's right.
Precisely to try to do something as academics to address this massive divergence
that you
keep putting up on.
Exactly.
Exactly.
So let me just summarize that with like a call to action.
Look, I think, you know, my sense is that as leaders, we need to understand and
actively
anticipate this change.
For me, this feels inevitable.
We won't have less internet, we won't have less data, we won't have less
connectivity,
we won't have less data.
Those genies are out of the bottles.
They're gone.
Yeah.
They're gone.
So embrace it and then anticipate, you know, hopefully we've given you some
beginnings of
a toolkit.
I want to think about how your economic environment and social environment is going
to get transformed,
right?
Drive innovation and institutional change across as a foundation for action, right?
That's going to be key, key, key.
Focus on becoming digital force and your endeavors, like what we're trying to do.
It's hard.
It's bloody hard for us, right?
But we have committed to this and because that's the only way we think we're going
to
have impact.
And scale.
And scale.
Exactly.
And so that's what we're doing.
You know, collaborate, right?
And then this setting is on competitive bottlenecks.
But make sure you just collaborate and work with people to solve problems, right?
You leverage partnerships and communities to drive, you know, your solutions,
right?
And then also think about the societal impact, right, and how that comes together.
And so with that, you know, I have a list of further readings.
That's great.
I love science fiction.
So there's a bunch of science fiction books that, for me, in form where the world
is headed.
So both Glenn Gibson, Ian Banks.
That's great.
You know, all these, all these authors and Lecky, all these authors really show you
what
might happen.
What might happen.
Yeah.
What the good and the bad.
Nice view of what to anticipate.
And then finally, you know, we were talking about Marco.
So this image we took almost three and a half years ago when our book first came
out.
So Marco was the ruins advisor.
Yes.
And also was the person that is my mentor and he's older than me.
He's older than me.
Yeah.
It looks that way.
You're nicely hiding your bald head with your hat because the middle of winter our
book
came out in January so we were doing a photo shoot and then it's a ruin.
And so anyway, it's so much fun to do to you.
It was.
It is.
It is.
I hope you enjoy the conversation.
Likewise.
Likewise.
Good.