Erik Brynjolfsson & Andrew McAfee: “The Second Machine Age” | Talks at Google


HAL VARIAN: Hi, I’m Hal Varian. I’m the chief economist
here at Google. It’s a great
pleasure to introduce Erik Brynjolfsson
and Andrew McAfee. I’ve known Erik for so
long, I can spell his name. [LAUGHTER] It’s really quite
an achievement. I haven’t known Andy that long. One F or two Fs? ANDREW MCAFEE: See? We’ll get there. HAL VARIAN: Well,
we’ll get there. Someday I’ll learn it,
but I read their book, and now I understand
all this effort that I put into spelling– it’s
not worth anything anymore. Two Ss, one F, all this stuff–
I just type a couple letters, and then some automatic
machine completes the spelling. So this is just the first
and simplest instance of what we’re all going
to encounter, according to the authors, so they’re
going to spend the next 45 or 50 minutes telling us about what
other jobs computers are going to take. Thank you. ERIK BRYNJOLFSSON:
Thank you, Hal. It’s a pleasure, and
according to our calculations, you will be the last person
to be replaced by a robot. Oh, you did make one
mistake in that intro, which is, we’re not going to
speak for 40 or 50 minutes. The reason we’re
here is, we want to hear a bit from you about all
the amazing things that you’re doing, and the
questions that you have, so Andy and I going
to try and speak for maybe for 15 or 20
minutes, and then do Q&A, and we’ll answer
whatever questions you have about the research
we’ve been doing at MIT, and the other companies
we’ve visited, and the people we’ve
interacted with, and will hopefully will be a very
lively dialogue that way. We realize that the best way
to learn about these issues is not to retreat to
our offices at MIT, and close the doors and windows,
but to get out and interact, and hopefully it’ll be something
that’ll be beneficial to you, as well. Now, the reason
we wrote this book was, in part, out
of our own confusion about a number of trends
that were going on, and they came from
different groups that we’ve been talking to. We visited a lot of companies,
like Google and technology companies in the Valley. We also spent a lot of time with
technologists at the Media Lab and CSAIL, the computer
science and the iLab at MIT and elsewhere, and
heard about the wondrous things that technology could do. We had a better experience
riding in the driverless car than Hal did. And we came away
from that with a lot of the infectious
enthusiasm and optimism that a lot of technologists
have about being able to solve all sorts of
problems with technology. But we also spent a lot
of time with other groups, in particular economists. I was just at the American
Economic Association meetings in Philadelphia, and it reminded
me why the economists are often called the dismal science,
because they weren’t nearly as optimistic as our
technology friends. They pointed out some stats
there weren’t so encouraging. Median income has basically
stagnated since the 1990s. The employment population
ratio has fallen dramatically, and hasn’t improved, although,
as you may have heard, unemployment has
gotten somewhat better. That’s mainly because more
people have dropped out of the labor force, not so
much because people are hiring. And there’s a whole
set of other statistics that aren’t as
encouraging, and so you contrast not just the
rapid innovation, but also some of the encouraging
economic stats, like the fact that we have hit,
last month, the record wealth in the United States– $77
trillion record GDP, record productivity, but yet these
numbers on median income and employment, and
it struck Andy and I as kind of a paradox, arguably
the great paradox of our era. Tracing the numbers
back a bit, we found that these numbers
weren’t always so decoupled– that through most
of the 20th century, and from what we can tell,
the 19th century as well, although the data
aren’t quite as good, these numbers all rose in
rose in tandem together. Productivity, employment,
median income– they were relatively
tightly coupled, but then starting about 15 years
ago, they became decoupled, with productivity and wealth
and aggregate continuing to grow significantly,
but the other numbers tailing off, or stagnating,
or even falling some. And this accounted for
these different perspectives that the technologist
or the techno-optimist had when they look at
one part of the story, versus some of the
economists, who were looking at another
part of the story. And we wanted to
try and reconcile how that could be true. I mean, was one
group right or wrong? And, in fact, they
were both right, but they were looking at
different parts of it, and furthermore, most
interestingly, we think that both
aspects of this– both the record productivity
and wealth, and the stagnating
median income– actually have a
common cause, and that is the fact that technology
has been racing ahead, but many people have not
been keeping up with that. There’s kind of a dirty
secret in economics that not many people
recognize, because it hasn’t been so important
before, and that is that while technology
can grow the economic pie, and really is what some
economists jokingly say is the only free lunch
that we believe in, there is no economic law
that says that everyone’s going to benefit from
those technological gains. It’s entirely possible that some
people will be made worse off. You’ve heard of the buggy
whip manufacturers that were put out of business
when Henry Ford introduced the automobile. Well, sure, there
are some people who may be made worse off. It turns out that the
size of that group can be almost arbitrarily large. It could be 50%
or more of people who don’t share
in those benefits, and the data suggests
that in the past 10, 15 years, that’s been the
pattern that’s been emerging is that technology has
been making the pie bigger, but most of those
benefits have been accruing to a
relatively small group. Maybe 60 to 70% of them have
been accruing to about 1% of the people in the
past decade or so, and a lot of that, while
there are many forces at work, a lot of it has to do with
the nature of technology. To give you an example,
consider a simple application like tax preparation. So TurboTax can do your taxes. Many of you probably
use TurboTax. I do. It’s a simple piece of software
that has codified that process, and reduced it to a basic
set of steps– digitized it– and once it’s become
digitized, guess what? You can make a copy of it. You could make 10 copies of it. You could make 100
million copies of it, and the marginal cost of
that is very, very low. Anything that’s digital can be
reproduced at nearly zero cost. What’s more, each
copy is virtually identical to the
original, and it could be transmitted
almost instantaneously anywhere on the planet. As Hal writes in his book,
“Information Rules,” which you should all read if
you haven’t already, these characteristics of
free, perfect, and instant are not common for most
other goods and services, to say the least, and they
lead to sometimes some unusual economics. Some of those economics
are what we call, winner take all markets,
because if someone has made a tax-preparation program, you
don’t necessarily want to go out and buy the hundredth-best
or the thousandth-best or the 10,000th-best program. You want to buy probably
the best one you can find. Maybe there’s some
room for a small amount of differentiation, if people
have slightly different kinds of needs, and so a very
small group– maybe one– will win the dominant
share of the market. Sticking with tax
preparation, that means that the people at Intuit
have done very, very well. There are some
billionaires there now. All of us, as consumers, have
done well, because for $39 we can get tax
preparation, and we have a lot of consumer surplus
and benefits– ease of use– that we wouldn’t
have had otherwise. But there are a bunch of
people in the tax-preparation business, at H&R
Block and elsewhere, that may have had a
college education, may have had some
skills that they’ve invested a long time in,
and now they’re not really adding a lot of value anymore. What they do isn’t
really all that useful to most of the rest of
us, and as a result, the ranks of people who say
their job is tax preparation has fallen dramatically over
the past five or six years. And it’s, of course,
not because taxes have gotten so much easier. It’s because technology
can now do that job. Now, that’s just one example,
but it’s, in many ways, a microcosm of what’s happening
in many other industries, where you can codify more
and more processes, and reduce them into a digital
form, and then replicate them. Marc Andreessen
has famously said that software is
eating the world, and he’s got a point
that this digitization is at the core of more
and more industries– not just the traditional
software industries, but, of course, music and
media, manufacturing, finance, retailing. In fact, it’s hard to
think of an industry that’s not affected, with the possible
exception of our industry, education, which I
think is next in line for that kind of
revolution, and that means that the same
kind of not just technology, but the
same kind of economics is coming to more
and more industries. A number of economists,
including our colleague, David Autor, and others
at MIT have a look at what’s happened to what they
call the skill content of jobs. You take the overall
occupations in the economy, and look at the underlying
skills required to do them, what you find is
that those jobs that involve routine
information-processing tasks, like, for instance, tax
preparation, but also travel agents, bookkeepers,
secretaries– those jobs have been the ones that have shrunk
the most over the past decade, because automation has
affected those tasks the most. Now, in doing the
research for this book, Andy and I spent time visiting
companies like Google, and we came to the
conclusion that it wasn’t just that
machines were getting good at routine
information-processing tasks, but they’re getting good at
a broader and broader set of tasks. I think Andy may be
talking about some of those in a little while,
but there’s a whole wider set of applications, and really,
in some ways, amazing things that machines can do
now, many of which I think people in this
room are probably inventing and discovering, and
that means that as disruptive as the past 10
years were– and it was a very disruptive decade– we
think that the next decade is going to be even more
disruptive, as it’s going to create even more wealth, even
more productivity, even more possibilities, but
also, it’s going to change the skill
content of more jobs. It’s going to mean
that more people will find that the existing
tasks they’re doing aren’t adding value to the
economy in the same way that they were before, because
they find themselves competing with machines,
and you don’t want to be competing with a $39 piece
of software if you’re human. Not many people want to work
for those kinds of wages. So in the end, do we come
away with this analysis as optimists, or
are we pessimists? Andy and I don’t put ourselves
in either the utopian category or the stagnationist category. We call ourselves
mindful optimists, because we believe
that we can address the challenges and the
opportunities created by these new technologies,
but that there’s no technological
determinism that’s going to automatically lead
to one future or another. The book is called “The
Second Machine Age,” but we learn a lot by looking
at the first machine age, which was the Industrial Revolution,
and in the Industrial Revolution, there were
some new technologies, some wondrous new technologies
that came along that bent the curve of human history,
that took us from a relatively stagnant previous
several millennia, in terms of living
standards, population, and many other metrics, towards
an explosion of productivity and increased living standards. And those technologies primarily
automated physical tasks– automated our muscles,
like the steam engine, and later, technologies like
the internal combustion engine. The second machine
age is much more about automating cognitive
tasks, mental work, and we think it’s going to have
equally profound implications. While, when you automate
make muscle tasks, it actually increases the
demand for control systems, or cognitive tasks,
and in that sense, humans tend to be complements
for the automated power systems. It’s not so clear that
automating cognitive tasks and augmenting cognitive tasks
is going to be complementary, versus a substitute,
for human labor. And as we work through
the implications of the second
machine age, we need to think about how we can
make the kinds of adjustments we made in the
first machine age, so that we don’t just
get the increased wealth and productivity,
but also the kind of shared prosperity that we had through
most of the past 200 years. So with that, let me turn
it over to Andy, who’s going to flesh some of
that out, and then we want to turn to all of you for
some questions and comments, and have a robust
discussion about this. ANDREW MCAFEE: Yeah,
and thank you, Erik. As Erik says, the main thing
we want to do is talk with you all, so I don’t want to
take up too much time. I want to do two things. I want to talk about
where this book came from, what its origin
was, and then what its goals became, as
we were writing it, because if anyone tells
you they know what they’re trying to accomplish
exactly with the book when they start writing it,
they have just lied to you. So I want to talk about what our
goals became, as we fleshed out this book that became
“The Second Machine Age.” I think if honestly
if there were one day where I had
this eureka moment, and where I realized that we
needed to go, first of all, get a lot smarter
ourselves, and then start writing and publicizing
it to the world, it would be in the fall of 2010,
when I’m reading the “New York Times” online, and there
is a story about the fact that this company developed a
car that was driving itselves on American roads in traffic,
with no human intervention whatsoever. And I spit my coffee at that
point for a very specific reason– this was not
supposed to happen. You all were not
supposed to do this. Shame on you. But but more broadly,
the self-driving car was supposed to remain firmly
in the realm of science fiction, and not over in the
realm of reality. We knew this because a
couple of colleagues of ours had written a fantastic
book six years earlier, in 2004, explaining, to
our complete satisfaction, why this driving a car
was going to remain over on the human side of
the balance sheet, and not over on
the digital side. And they made a very
convincing argument that the pattern-matching
work that had to go on– high-stakes
pattern-matching to drive a car. If you get it wrong, you’re
going to hurt somebody. The amount of
information you had to take in, the different
sensory channels you had to incorporate, the formal
and the informal rules of the road that you had
to be aware of– this was a daunting task. We humans are kind
of good at it. Technology was lousy
at it, and that was the way it was
going to remain, and Erik and I
both read the book. We taught it in our
courses, we believed it, and then you all
screwed up the equation by basically building a car
that did this kind of stuff. And then over and over
again, we saw these examples of science fiction
becoming reality, and a quote from Hemingway
kept echoing in my head. He has a wonderful quote
about how a man goes broke. He says it’s gradually
and then suddenly. And we kept on seeing these
examples of really incremental, really kind of boring progress,
turning into holy cow, we’re done, eureka, this
mission is accomplished. How many people have
heard of the problem in robotics of SLAM? I just want to see how
geeky this audience is. Oh, fantastic. I know something more
than a group at Google. I’m going to go
home happy today. SLAM is another thing that
we humans do really easily. Everyone, point
to the staircase. Duh. All of us could locate
where we are in the room right now, because
we engage in what’s called Simultaneous
Localization and Mapping. All of us know where the
staircase is in the room. All of us know where the
doors and the hallways are. We know where we
are in that room. It turns out it’s fiendishly
difficult, astonishingly difficult, to give a robot that
kind of capability, as well. There is a review of
the literature on SLAM that was published, I
believe, in 2008 that said, we’ve made essentially
no headway, and we can’t see what’s going
to change going forward. This is one of the holy grails
of the robotics discipline. A colleague of ours named
John Leonard, back at MIT, last year, or no more
than a couple years ago, essentially solved SLAM for
a medium-sized room– I’m not making this up– by
waving a Kinect around. So he took $150 of
consumer equipment, and basically solved this
long-standing challenge in robotics. So over and over again,
we kept on seeing these really nasty, thorny
problems just being blown away by smart geeks working with
incredibly powerful computers and oceans of data, and that
was really the motivation to write the book– for,
first of all, for me and Erik to understand this world,
and then to communicate it. And then two other
goals– the second thing I want to talk about
is the goals that appeared as we’re writing the
book– two other goals became apparent. The first was that
we felt like we had to do battle with
two forces out there, and Erik has talked about them. One is the force of,
there’s nothing new here. There’s nothing to see here. Move on. Technology has the same
role in the economy that it’s always
had in the economy. It’s not going to lead to any
bigger or different disruptions than has gone on before. I just don’t believe
that, and I felt was important to try
to stake out that turf, and defend the claim
that honestly, this is some kind of new regime. This as a second
machine age here. The second one is kind of
the opposite of the nothing to see here. It’s the nothing to worry
about here, because everything is going to be
hunky-dory because of technological progress. We need some utopians out there
who say, oh, you and your silly worries about jobs and wages
and workers and all that. We’re creating a
world that’s going to be so munificent that
essentially, everything’s going to be free. No one’s going to have a job. We can all just do Maker
Faire all the time, and things are going
to be fantastic. So what are you boring,
Northeast, academic people worried about? Just come hang out at
Food Camp for long enough, and you’re going to walk
away just totally optimistic about the state of the world. We don’t buy that, either. There are some real
challenges coming up because of this technological
progress, and to ignore them is a pretty bad idea. As Erik said, right
now it’s pretty clear that there’s some polarization
and some hollowing out going on in the middle class of America,
because the stereotypical, median, American worker does
not have a college education. Fewer than 40% of us do, and
is not doing manual labor. They’re doing low-level,
fairly routine knowledge work. As with most of us
know, that is squarely in the sights of this wave of
progress that we’re seeing, but that wave
doesn’t stop there. Most of us would not call
a doctor a median worker. Most of us would not
call medical diagnosis a routine
information-processing task, but what Watson is doing–
since it’s won “Jeopardy,” it needs a new job– it
has gone to medical school, and I’m pretty convinced
that if it is not today the world’s best diagnostician,
it will be fairly quickly. So these disruptions
in the labor force are not going to be confined
to that fairly low-wage, fairly average worker. They’re going to go both
up– up in the age, the wage, and the skill, and the
education distribution– and then downward, as well,
because we’re doing things like learning how to pick fruit
automatically, which is not a prestigious job, but
certainly a human one now, too. And then our last
goal was to try to offer some prescriptions
about what to do, and I want to be pretty clear. Our prescriptions
are not out there in the singularity future
kind of time frame. Ray Kurzweil has the 2045
turf staked out pretty well. We don’t want to go
encroach on that. We don’t know if
we’re going to get to his version of the future. One of the funniest things that
ever happened to me, as we were writing this book– I
was on stage with a very prominent roboticist, who
I don’t think I’ll name, and he got asked a question
about the singularity. And he gave a response
that I thought was incredibly rude,
until I thought that was incredibly accurate. He said, look, Ray
Kurzweil is going to die. [LAUGHTER] All right. So we don’t want to be out there
in the 2045 time framework. What Erik and I are
talking about in the book, and our recommendations,
focus on time frames a lot, a lot closer in than that,
and I can bucket them into two time phases– things
we could and should do tomorrow, the immediate stuff, and then
the slightly longer-term stuff. The immediate stuff
is what you would learn if you pick any Econ
101 textbook off the shelf. I don’t care if it was
written by Greg Mankiw, who advises Republican candidates,
or Paul Krugman, who’s a pretty well-known
liberal economist. Their Econ 101 textbook
would tell a government to do a small set of
things, and to do them very, very well– to do education,
to foster entrepreneurship in an environment, to
double down and make sure you’ve got
world-class infrastructure, and to have a fairly
welcoming immigration policy, and to invest in
original research. This is uncontroversial
among economists. We think it’s exactly the right
the immediate-term playbook to restart economic growth, and
therefore, restart job and wage growth, as well. Looking out slightly
longer term, the crazy, radical ideas
that we have in the book are things like– how many of us
have heard of the Earned Income Tax Credit? Good. This is basically a negative
income tax for people lower down on the wage
and income distribution. It rewards a dollar of
work with more than $1 via the income-tax system. Great idea. Let’s subsidize
labor, especially at the entry points
of the workforce, as heavily as we can. Milton Friedman, that
notorious socialist, advocated for a negative
income tax a lot. We think that’s
a fantastic idea, so we believe there are more
than tweaks– interventions you can do in things
like the tax code that will go a long way toward
putting people back to work. Farther out, again, we
are not singulartarians. We’re not talking about the
economy of 50 years from now. For what it’s worth, I think
it’s a science-fiction economy. I think it’s, we’re
at the holodeck. I think we’re in
some crazy place, because you all
are going to keep doing what you do for
the next 50 years. It’s going to take us into a
world that I can’t envision. What I do know, though,
is it could take us into a more utopian, kind
of a cool, “Star Trek,” explore new worlds. The phrase we use when
we’re not talking to geeks is a digital Athens for that
world, where the citizens have this enlightened
discourse all the time, and instead of being
supported by human slaves, they’re supported by
a big army of machines that do the hard, dirty,
heavy lifting in an economy. There are dystopian
strains, as well, and if we continue to
polarize the economy, if we provide no opportunities,
if mobility continues to go down– those are
outcomes that I think push us in more of a
dystopian direction. The broadest point I think that
we make when we write the book is that the choices that
we all make, starting now, are going to determine
which of those two paths we create when we head
down, going forward. I think that’s probably
a good place to shut up. Let’s turn this open, and see
what people want to talk about. You’ve had your hand
up for a little while. AUDIENCE: Yeah, I
was just wondering– you were talking about
negative income taxes, and one thing that
comes to my mind is the patent-protection
system, where it essentially grants government-sanctioned
monopoly to the person doing the R&D. If technology
improves, would you advocate some sort of reduction
in the patent window, because that would increase
competition, and then share? ANDREW MCAFEE: Yeah, this is
an amazingly controversial area of work among our colleagues–
IP log and IP regimes, in general. The stuff that makes
the most sense to me, and I want to hear
what Erik says, are that we’ve
probably gone too far with some categories
of IP protection. It’s not clear to me
that software patents are doing more good than harm. We might want to revisit that. Companies like yours buy
these companies mainly as a war chest for
patents, and then you go rattle the saber
against some other company, and then you have some
kind of agreement. That’s not value-added activity. Amazingly enough,
there’s a cartoon made by Disney called
“Steamboat Willie” that is still under copyright. I think it was
made in the 1920s. It’s the precursor to
Mickey Mouse, which Disney wants to hold
onto tenaciously. I’m not sure what purpose
is served by that. So I’m a huge fan of
property, in general, but I think our property
protections might have gone a bit too
far in the IP world. Does that make sense? ERIK BRYNJOLFSSON: Yeah, I
think you got it exactly right, and I would just
elaborate by saying that the purpose of
intellectual property, according to the Constitution,
and for that matter, according to economists, which,
in this case, happen to agree, is to foster the creative
arts, to foster R&D, and I think, in
many cases, we’ve lost sight of that
with a regime. It’s been taken
over by lobbyists. The copyright of Mickey
Mouse, and the fact it’s been extended
and re-extended– there’s the Sonny Bono
Copyright Extension Act– that basically retroactively
extended copyrights. It’s hard to come up
with any rationale how that would increase the
incentive of people in 1920 to make more inventions
by giving more wealth to the people who
own those copyrights today. So that’s a perfect–
and then software patents is another one where, when you
build on previous innovations, then patents can
discourage innovation, rather than
providing incentives, because they basically take away
the returns from the people who are building the new
stuff, and giving it back to people who have built
the pre-existing stuff. So for all those
reasons, there’s some serious re-thinking
that needs to be done, or actually, returning
to the old thinking that needs to be done. AUDIENCE: What do you think
are the greatest risks to your predictions about
rising unemployment and falling median income and hollowing
out of the middle class and so forth? ERIK BRYNJOLFSSON:
Well, just to be clear, we’re not necessarily
predicting that. We see that as something that’s
happened in the past few years, and as Andy said
at the end, and I tried to say at the end, as
well, the outcomes are going to be very much dependent on
our choices, and when you say, what are the risks,
do you mean in terms of risks of us getting it wrong,
or risks that if that happens, what will be the downside? There’s two interpretations
of the word, risk. Which one do you
like us to address? AUDIENCE: What are
the factors most likely to prevent
those things happening? ANDREW MCAFEE: What can we do if
we don’t like the trajectories? What can we do? AUDIENCE: Well, it’s partly that
kind of proactive what can we do, but it might also
be, trends play out in a way very different– ERIK BRYNJOLFSSON:
Yeah, yeah, yeah– AUDIENCE: –than
you would assume. ERIK BRYNJOLFSSON: –and we
would like to see that happen. I mean, we can learn a little
bit from the first machine age, because that was also a
time of great disruption. It was almost exactly 200
years ago this year– actually, it was– that the
Luddites started smashing the looms in
the machines in England because they saw them
as taking away jobs. And they had a point
that there were a lot of people– weavers–
who were losing their jobs, but at the end, new jobs
were created by the economy, and you could argue, and it’d
be a fact, that technology has always been destroying
jobs, and it’s always been creating jobs, and over
time, historically we’ve created as many new jobs
as we’ve eliminated. Now, those trends have become
decoupled in the past 15 years. If you look historically,
one of the reasons that we did create new jobs
was because we, as individuals, as entrepreneurs, and
as governments, we made a whole series
of choices, which helped maintain
shared prosperity. Just to give you a
couple of examples– it’s been called
America’s best idea– is universal, mandatory
education, and that helped bring a
bunch of people who used to be farmers, and give
them a set of skills that made them suitable for a whole
set of other types of jobs and occupations, and
America led the world in pushing that, and
not coincidentally, led the world in shared
prosperity and living standards. But it wasn’t just
some government action. It was also a bunch
of entrepreneurs that helped invent and
discover the new industries. 90% of Americans used
to work in agriculture. Now it’s less than 2%. That difference didn’t join
the ranks of unemployed, thankfully. They were employed in
totally new industries that people like Henry Ford
and Steve Jobs and Bill Gates and Larry Page and Sergey
Brin invented that people back then probably could never
have even conceived of, so it’s very much the
role of entrepreneurs to help identify what
those new industries are. The data say that right now,
we aren’t doing it as fast as we used to. We’d like to see
more entrepreneurship to help speed up the creative
destruction, especially the creative side. Surprisingly, when we
looked at the data, we found that in
the 2000s, there were fewer new enterprises
and job creation than there were in
the ’90s or the ’80s, so we think of this as being
a very entrepreneurial era, but actually, the data suggests
that in some ways, it’s not. If we could speed that up,
if we could, in some sense, crowdsource this
question of where will the jobs come from,
which is historically how we’ve solved the
problem, we think we’d be hopeful that we could
improve on that situation. But we have to say that in
fairness, we are not completely sure that the future
is going to play out just the way the past did. It’s been said– I heard
this from Hal Varian– that history doesn’t
repeat itself, but sometimes it rhymes. I’m not sure where you
got that from, but– ANDREW MCAFEE: Hal Varian looks
a lot like Mark Twain, I think. ERIK BRYNJOLFSSON: Yeah,
yeah, I think maybe that’s where he got it from. So maybe we will be
able to replicate some of the successes,
but I’m sure it’s going to require some
different– we can’t just double down on doing the exact
same thing we did before. We’re going to have to have some
different kinds of policies, and that’s why we wanted to
start this conversation was to help identify what
those new changes are going to need to be. ANDREW MCAFEE: I like
your question a lot. I think the main thing that
might make the future play out differently than I am
envisioning right now is if Erik and I made the
classic geek mistake, when we were writing the
book, of becoming too enamored of the technologies
we were looking at, and projecting them
too far in the future, and thinking that their
impact is actually going to be bigger
than they are. I don’t think we’ve made
that mistake, but if we have, then economic growth will
have some really different consequences, and the
employment impact of technology will be less than I am
currently thinking it is. That’d be kind of a
good mistake to make. AUDIENCE: So a little
more than a year ago, Paul Krugman wrote
an op-ed in which he observed that a lot
of the new technologies are going to start putting more
wealth in the hands of people that have capital start
with– people that can afford the robot army that then
takes away manufacturing jobs. So I’m wondering what kind
of economic restructuring do you need to go through
to be able to deal with that new paradigm? It almost happened
before, like you’ve mentioned, in the
first machine age, but it seems like it’s going
to be on a lot larger scale this time. ANDREW MCAFEE: The great
phrase that Krugman used was “robots and robber
barons,” right, are going to get all the
benefits of the world that we’re creating,
and he thinks that’s a pretty bad outcome. We do, too. What can we do to make
sure that doesn’t happen? Again, we’re going to sound a
little bit like broken records. What we don’t want to do, what
we’re really, really careful– what we’re sure we
don’t want to do is storm the barricades
of capitalism, and fundamentally rethink the
economic contract right now, in 2014. Most of it’s
working pretty well. The biggest problem
that you identify, and that we spend time on, is
it might get harder and harder to make a living
from your labor, as opposed to from your capital. If that’s really what’s going
on, let’s remove the barriers. Let’s try to upscale our labor. Let’s remove any tax or whatever
distortions to doing that, and let’s see how that goes
before we start thinking about getting rid of
capitalism or demonizing capitalists or things like that. AUDIENCE: So if you
assume that there’s something called structural
unemployment, which, for my purpose of my
question, means someone whose potential benefits
of being employed are less than the costs
of employing them. I guess my question is,
do you think that exists? Do you think there’s
a population that fits that category? How large do you
think it is today, and what do you think
is going to happen to that number over the
next coming decades? ANDREW MCAFEE: Sir? ERIK BRYNJOLFSSON: Sure. There are a number
of ways that you can get– the
answer is yes, I can see there is some
structural unemployment, and there are a number of
ways you can get to that, and some of them are
very discouraging. If you look to– in the 1930s,
when much of agriculture– when tractors were
introduced, mechanizing a lot of agriculture, Joe
Stiglitz and others have pointed out that
took tens of millions of people who used
to work on farms, and threw them into
the labor force, and it caused a
macroeconomic, or helped contribute to a macroeconomic
recession, or the Great Depression, and it took a
decade or more for many of them to be reskilled, and
for new industries to crop up to hire them. And you can easily see
that that kind of mechanism could happen today, not with
farmers, because there aren’t any of them left,
hardly– there’s only 2% of the population
and falling– but in, say, routine
information-processing tasks. And one of the things that
concerns me is that by the time you find new skills and new
jobs for those people to do, what if the technology
has changed again? What if you have
another disruption, and then you’re in this constant
pattern of trying to catch up with what’s going on before? And if the pace of
technology speeds up, then you could see
that not only would you have structural
unemployment, but it could get more and
more severe over time. One of the ways this
manifests itself is not just through unemployment,
but also through inequality. If you’re a true
free-market economist, you say, well, markets
will always clear. I mean, you just have let
them clear, and at some level, that’s what the math says. That’s true, but the
wage at which they clear, the price at which they clear,
could be arbitrarily low. It could be $5 an hour. It could be $0.05 an hour. It could be five-hundreth
of a cent an hour. And I think most of us would
agree that at least two of those numbers, and
probably all three of them, aren’t ones we feel very
comfortable as seeing as the clearing wage for labor. AUDIENCE: It could be negative. ERIK BRYNJOLFSSON:
Well, if you figure the cost of hiring somebody–
like right now, if you look at– somebody gave me
a horse, and said here, you can have a horse, and you
can use it to ride to work and whatever else you
want to do with it, I’d say, thanks, but no thanks. I really, even if I got for
free, it wouldn’t be worth it. And so you’re certainly right
that an input of production, once you cover the other
costs associated with it, even 0 might be
too high a price. ANDREW MCAFEE: But the question
that Erik and I have wrestled with the hardest, as we wrote
this book– we’ve had not just bantering, but honest arguments
and disagreements about it– is, are we heading
into an area where the average worker is a horse? In other words, I couldn’t walk
up to a factory or a warehouse or an office park with
a horse and say, hey, you can have this horse for
a tenth of a penny a day. Put it to work. They’d say, get lost. I have no use for that input
in my system of production anymore. We’re not headed there
tomorrow or next year or next five years, even with
this ridiculous technological progress. The question I honestly
don’t know the answer to is, are some of us heading into
horse territory in the economy that we’re creating? Most of the really
smart economists that we talk to find that a
fairly ridiculous proposition. I was talking earlier
this week with a guy who’s got a Nobel Prize, and I
said, does your intuition tell you that if we
play our cards right– and that’s not a
given, obviously, with the polarization
in Washington and everything– if we
play our cards right, though, that we can be back
to a full-employment economy after this nasty transition
maybe lasting 20 years? And he said very confidently,
oh, yeah, absolutely. We can get back to a
full-employment economy. I just find myself less
confident than that. And again, I want to
stress, it’s your fault. [LAUGHTER] AUDIENCE: There’s no doubt
that technology is contributing to the hollowing out
of the middle class, but there’s another effect,
which is technologically related, and I’m
not sure if you’re talking about it
directly or indirectly, but just the fact that
communication barriers have come down, and so now a
lot of knowledge work that used to be done in America
can now be done elsewhere, which is good for
other countries, but just not so
great for this one. So how do those two
variables interact, and what will the trend be? ERIK BRYNJOLFSSON:
Well, you’re right. I think that there are
two big trends that are affecting the
economy today– globalization and technology. I mean, there are some
other, smaller trends, in terms of cultural
changes, unionization, but just focusing on
those two big ones, I agree entirely that
globalization is, in a large part, enabled by
some advances in technology, especially communications
technologies, also some transportation
technologies. But of the two, we think
that the technology one is the one with more legs, and
the one that’s accelerating, and if you look
at a lot of kinds of jobs that have been offshored
historically, now many of them are coming back to
the United States. Which low-wage
country do you think Apple, on their
30th anniversary, is making their
Mac Pro computers? MALE SPEAKER: USA. ERIK BRYNJOLFSSON: USA. They’re making them
in Austin, Texas, not because the wages
are so low in Texas– maybe we should say,
not just because they’re so low in Texas– but not
because of the wages there, but because there’s hardly
any labor required anymore. They’re bringing them
back here, and that’s happening for a lot of
knowledge work, as well. Larry Summers pointed
out to us once when we were talking
about this that if you look at manufacturing
employment in China, how has that changed in
the past 15, 20 years, as manufacturing
has grown in China? It’s actually fallen. There are 20
million fewer people working in manufacturing now
than there were 20 years ago, so manufacturing employment is
leaving both the United States and China, not going to
Vietnam, but going to robots. And so, in many ways, our
view is that if anything, offshoring is just
a weigh station on the road to
automation, and some of these low-wage
countries are even more in the bulls-eye of automation
in this second machine age than American or Western
European or Japanese workers, because the more routine the
tasks are that you’re doing, the easier it is for someone
to write a piece of software, or a robot or a
program to do that. And if your competitive
advantage is, hey, we’ll do it for
cheaper, well, I don’t think that’s a very
strong barrier to entry, compared to what’s
happening with the pace of technological change. ANDREW MCAFEE: And
you asked specifically about knowledge work. I came across this really
interesting series of stories in the press awhile
back, because some different
reporters noticed it, and then they were calling
up their insurance company or their credit card
company or their bank. They seem to be interacting
with this perfectly accented American voice that was
giving them somewhat canned responses, with the same
sentences or sentence fragments over and over again. And so they pushed
on that, and they noticed that they think there
are two different strains going on right now. One is where there’s a Filipino
listening to what they’re saying, because English
skills are pretty good there, and then pressing
a button for which American-accented
response they’re supposed to spit
back on the fly. The slightly more
advanced version of that is there’s an algorithm
listening to what the person is saying, and the
algorithm hits the button for which pre-recorded human
voice should be given back to the person. Step three of that
progression, I believe, is– we have all seen
the movie “Her”– is Scarlett Johansson’s
synthetic voice is just going to give us a response back. ERIK BRYNJOLFSSON: That’s the
future we’re all hoping for. ANDREW MCAFEE: Yeah, exactly,
can’t get here soon enough. [LAUGHTER] ANDREW MCAFEE: But it’s not
science fiction stretched anymore to imagine a completely
synthetic interaction like that. Again, for knowledge
work, we think that offshoring
is a weigh station on the way to automation. AUDIENCE: How much
do you think this is a US issue, versus
a global issue? ERIK BRYNJOLFSSON: Well, it’s
certainly a global issue. I mean, if you look
at the– I mean, a lot of the trends we gave
were using US numbers, frankly, because we are the most
familiar with them, and some of the best
data is available here, but if you look at the OECD,
the Organization of Economic Cooperation and Development,
the most developed countries, you see a very similar pattern. For instance, the hollowing
out of routine jobs has happened in virtually
every one of those industries, according to research by
my colleague, David Autor, and others. The inequality has grown
in all those countries. It has grown in Sweden,
Germany, France, Japan. There’s only two
or three countries where it hasn’t grown– Greece
and, I think, one other one, and there’s just
some weird things going on in those countries
that probably aren’t typical. And as we just said, when you
look at the developing nations, you see, I think
arguably, that they may be even more
in the bulls-eye, and I think this could be a real
concern, because historically, having low wages and taking
some of those kinds of jobs in textiles and other areas
has been a path to development to get out of that
underdeveloped state. You for awhile do those
really low-wage jobs, and then, as you work
your way up the ladder, you start joining
the broader economy. If that path is cut off,
it could be very troubling. Those countries that
haven’t made the transition may be stuck for a long time on
the other side of the ledger. ANDREW MCAFEE: There
was a nice piece of work done by a couple
of University of Chicago economists just recently
that looked at the returns to capital versus labor in many
countries around the world, and they found the pattern
was really similar, almost without exception. The share of total GDP that
was getting paid out in labor was heading south country
after country after country around the world. We can think of an
explanation for that doesn’t involve technology. And in fact, they said the
explanation is explicitly information technology. ERIK BRYNJOLFSSON:
Right, and I think that that’s a very important
point to emphasize, is because there are
people who say well, it’s because of what
their Congress did, or what the guys in Wall
Street did, or whatever, and I’m sure there have been
a lot of bad things done in Washington and Wall
Street and elsewhere, but the fact that this pattern
is so pervasive and so global suggests to us, or is
additional evidence to us, that there are some more
fundamental economic forces and technological
forces at work. AUDIENCE: I think once you
have big problems like health, and legal system politics
become automated, like having a differential
diagnosis done by a machine or a
phone app, and it tells you exactly
what’s happening, you probably have
more trust of a better diagnosis than the
best doctor out there, and for this layer of very
high-paid cognitive type of skills, there is a lot
of infrastructure built to protect this system, and a
lot of politics and lobbyists that want to make
sure that it stays. And can we create
a baseline for, let’s say, before we
go that route, there needs to be a unanimous approval
on what can go– not to serve a particular interest of a
certain group, but rather, humanity at large? ANDREW MCAFEE: So no is
the short answer to that. ERIK BRYNJOLFSSON: After that
polite way of [INAUDIBLE]. [INTERPOSING VOICES] ANDREW MCAFEE: Yeah, it
was a lovely question. I’m going to dismissive answer. No, we certainly can’t change
our political structure that deeply, so we have
to have a plebiscite before every instance
of technological change and lobbying protect– we’re
not going to get there. But you bring up a
really important point, which is that– I
won’t speak for Erik– I think that the main impediment
to a lot of the stuff, the great stuff, that
we’re going to see is exactly the barriers
that you identify– our regulatory barriers that
are there to protect incumbents, as opposed to protecting
consumers– they primarily have that effect– are
lobbying groups or all the forces of inertia
that are trying to keep stuff from happening. When you look at the– some
of Uber’s recent troubles are self-inflicted. Let’s be really
clear about that. Some of the other
ones have been brought on by exactly the
forces you identify. Did you all hear
that France is making Uber wait a minimum
of 15 minutes before the car can show up
purely to protect their taxis. This is insane, but we’re
going to see all kinds of stuff like that go on. Stupid question
time– how many of us have heard of Tim O’Reilly? Heh heh. Dumb, right? Tim has a beautiful
way to phrase it, and it’s something
that I carry around. He says, look, with our policy
and our regulatory decisions, we have two choices. We can protect the
past from the future, or we can protect the
future from the past. For heaven’s sake, let’s protect
the future from the past. ERIK BRYNJOLFSSON:
Right, and that’s not to say that there aren’t
policies that we should adopt, but I 100% agree with Andy
that there’s a natural reaction to either smash the machines,
the Luddite approach, or to freeze the existing
occupations and jobs the way they are through a whole set
of occupational licensing and another incumbent
protection acts. And while those seem
like natural ways to prevent some of
the bad outcomes, ultimately we
think they’re going to be more destructive
than beneficial. The way that the economy
has thrived in the past has been what does
Joseph Schumpeter called creative destruction,
and the answer is not to try to slow
down the new ideas and the new technologies. It’s to speed up our
adaptation, and the government can help in many ways by, for
instance, I mentioned earlier, universal public education was
one idea, and raising the skill levels. Encouraging, boosting, and
speeding entrepreneurship– not slowing it down. Doing things with tax
policy, doing things with R&D. So there
are lots of things you can do to help
with the preconditions to smooth and speed
the transition, but freezing in the
historical structure is certainly going to
do more harm than good. AUDIENCE: Hi, so
companies like Apple make money by building fancy
gadgets to sell to consumers, and companies like Google make
money by allowing advertisers to advertise their fancy gadgets
to consumers, and all of this requires that the consumers
are there, and they have jobs, and they have– MALE SPEAKER 2: Money. AUDIENCE: –money. So in a sense, we
have an interest in making sure that this
situation you talked about where workers turn into
horses doesn’t actually happen. Are there any economic
models which talk about this, and say that there might
be some kind of tipping point where it all gets into
a horrible, negative feedback loop, and? ANDREW MCAFEE: We can we
can summarize the situation you’re describing, in
an apocryphal story about Henry Ford II
and Walter Reuther, who was the head of the
auto workers’ union. There’s a story
about the two of them going through a very highly
automated, brand new plant that Ford had
built, and they were kind of joshing with each other. I’m sure they hated
each other deeply, but they were kind of
joshing with each other, and Henry Ford points to one
of his new pieces of equipment, and he says, hey,
Walter, how are you going to get those
robots to pay union dues? And Reuther says, hey,
Henry, how are you going to get them to buy cars? So this is exactly the problem
but that you’re identifying. If people are out of work, if
their wages are going down, aggregate demand in
the economy goes down, we know that’s a really
bad thing to have happen, and what I don’t know is
how soon is this coming? How bad will it be? But any reluctance
to spend is going to cause the economy
and overall demand to head in the wrong direction,
and that reluctance to spend can be based in
reality, or it can be based on Keynes’
famous animal spirits. ERIK BRYNJOLFSSON:
Right, and the example I gave earlier about what
happened in the Great Depression– Joe
Stiglitz described that as being one of the contributing
factors to that fall in demand, and so it’s going to potentially
have macroeconomic effects, as well as these
distributional effects. AUDIENCE: I wanted
to ask you something personal and specific. ANDREW MCAFEE: And I see
we’re just about out of time. [LAUGHTER] AUDIENCE: Last week there
was a lengthy article in the “Economist” about
the future of jobs, and it liberally
quoted from you– ANDREW MCAFEE: Not liberally
enough, I’ll tell you that. AUDIENCE: –in your
upcoming book, and so I wanted to ask you if you are a
parent, or want to be a parent, what would you
advise your children to do who are going to enter
the workforce in the next two decades? What careers and what kind
of education opportunities should they pursue, because
the classic types of things that maybe we have told them in
the past– go become a lawyer, go become an accountant,
go become a doctor. This is a ticket to
professionalism, upper middle class lifestyle, and the odds
of you becoming unemployed are very low, and I
don’t feel that that is going to be the case
going forward anymore, and so I want to
put that on you, and if you had to
advise your own son or daughter, what would you say? ANDREW MCAFEE: Let’s have
the parent answer first. ERIK BRYNJOLFSSON: It’s a good
question, and it’s a tough one, but there are still
some categories that machines are
not very good at, and we are hesitate to–
we almost never say never because of what we’ve seen
in terms of advances– but things that involve
significant amounts of creativity, invention,
entrepreneurship. Those have not been
affected at all. In fact, they’ve been augmented. They’ve become more valuable. Interpersonal interactions
and skills– machines so far are still not very
good, although we’ve got friends in the Media Lab
and elsewhere that are there pushing the frontiers on that,
but whether that’s motivating people, or even just
caring for a sick child, comforting somebody–
that’s something that machines don’t
seem to be very good at. I would encourage my
kids to be flexible, because whatever types of tasks
that are or aren’t automated now– it’s likely to
change, and people are going to have
to change and adapt much more quickly
than in the past. So we don’t know for
sure exactly which categories are going
to be affected. Those are some of them, but
maintaining that flexibility is probably going to be one of
the key things you can do you. ANDREW MCAFEE: You asked
a personal question. I’ll give a little bit
of a personal answer. I’m not a parent, but
the most common question that we get asked
about this stuff comes from parents
of kids of all ages saying, what advice do I give? What do I do? So my advice for
parents of young kids is actually a lot easier for me. I say, just send
them to Montessori. And not that that’s a panacea. Are there Montessori
kids in the room? Really? That’s amazing. If one or both your
founders were in the room, hands would be going up. There’s a Montessori mafia
among the technology elite, and I was a Montessori kid, too. And what it taught
me is that the world is a really interesting
place, and your job is to go poke at it. I think that’s great. My school only went up
through the third grade, so in the fourth grade, I went
into the public school system in the Indiana town
where I’m from, and I felt like I had
been sent to the Gulag. I’m like, you want me
to sit here all day, and a series of things
is going to be inflicted on me at 45-minute intervals? What are you people, sadists? It just made no sense
to me, whatsoever. So early on, go teach
your kids that the world is an interesting place. At the college level, my
silly bumper sticker advice is put down the
beer pong paddle, and just go do a double major. Go hang out with the theater
geeks at one side of campus, and the math geeks at
the other side of campus. Fill up your toolkit
as much as you can. ERIK BRYNJOLFSSON: And the only
other thing I would add to that is that if we do tend to more
of a winner take all market, there’s even more of a premium
than there was in the past on being the very
best in something, so it becomes important
to pursue things that you’re really
passionate about, that you’re interested
in, because we all know that you’re not likely
to be the best at anything that you’re just doing
because you have to or because you’re supposed to. But there are going to
be niches– large niches and small niches and tiny
ones– where somebody has the potential to be
among the best in the world, and those are the
ones that are still going to be in demand, whereas
if you just another person doing a job because
you have to, it’s going to be increasingly
hard for people to find a way to make
a living doing that. AUDIENCE: We seem to
take it for granted that investing education
is a wise thing to do, at least on a
macroeconomic scale, but on a personal
scale, given the rising popularity and prosperity,
if I’m not rich, and my prospect of becoming
well-off are very, very slim, why should I be motivated to
invest in educating myself if the chances of that having
me better off are so low? ERIK BRYNJOLFSSON:
Well, the data suggests that even today,
the difference in earnings between people with more
education and less education is enormous, and
it’s been growing, so from a purely selfish,
personal point of view, it does pay to get
more education. But it’s important
to understand, when we talk about
education, first off, we don’t just mean college
or post-graduate education. We mean all levels, whether
it’s K through 12, vocational, lifelong learning– all of them
building more human capital in any of those dimensions
is likely to add value. But we also think that education
has to be reinvented exactly along the lines that Andy
was describing earlier. The traditional pattern of
having people sit in rows, and learn to follow
instructions carefully from the person at the
front of the room– well, following instructions is
something that machines are really pretty good at doing,
so education of that model, and imparting facts
and numbers, isn’t likely to be the
kind of thing that we need more of the future. Education that
sparks creativity, that allows you to think about
new ways of putting things together, and create
things– that’s much more likely to be something
that adds value. ANDREW MCAFEE: Here, here. AUDIENCE: Thanks for
a wonderful talk. So coming back to
this policy question, how aware would you
say the average policy maker or politician in
US is about these trends, and how could we actually
get more of our politicians to understand the nuance of
what you’re talking about? ANDREW MCAFEE: Someone needs
to write a book about it. [LAUGHTER] ANDREW MCAFEE: You can either be
really pessimistic or somewhat optimistic about it. I’d say the awareness is early,
early days, but growing kind of quickly, and everyone
that we talk to, these really well-informed
people, bemoan the polarization in Washington,
and this environment we have where, if you say
black, by definition, I’m going to say white. But the green shoots
that we see are around some pretty important
areas, like immigration policy. There is a pretty broad
bipartisan agreement that we need to
change some things. There’s a lot of talk about how
we need to reinvent education. No one’s exactly sure
how we need to do that. So I would say the awareness
is, especially about this tech stuff, because Washington’s
not a technophilic place, the awareness is low. It’s getting better
kind of quickly, and maybe it’ll get
OK at some point. ERIK BRYNJOLFSSON:
Yeah, I mean, that’s certainly one of the
more depressing parts of our analysis is
just the way things are so stagnant in Washington and
in public policy, in general. There is a growing
amount of anger, whether it’s the Tea Party
or Occupy Wall Street, or other groups, but our sense
is that, by and large, they mostly have misdiagnosed
the nature of the problem, and if you don’t have
the right diagnosis, you’re probably not going to
have the right prescriptions. And so a big part of the reason
we wrote this book was to help change the conversation. Ultimately we’ve
been told by people who know better– Washington
doesn’t lead the country. It’s people who
lead it, and they will respond to what
people are demanding. So phase one is
spreading the awareness, as I think you just
brought up, getting people to understand some of the
issues that are important, and once that conversation
has been changed, we’re going to have much
better hope of changing the kinds of policies. AUDIENCE: So you
talked quite a bit about what regulators
can do to counteract some of these effects. Is anything we can do,
given that at Google we are pushing technology forward
and are often techno-optimists? ANDREW MCAFEE: Yes, stop, OK? ERIK BRYNJOLFSSON: No, no, no. ANDREW MCAFEE: No, don’t. That’s the first thing. This is actually
really important. We come across a lot of
really well-meaning people who kind of want you guys
to stop what you’re doing, and want the tech sector
to slow down and stop putting people out of work. We think that’s the single-worst
idea that we come across, so by all means, carry on
taking us into the future. It’s incredibly important work. As individuals, advocate,
support enlightened policies, vote for and support people
who are taking our policies in the right direction, and just
participate in the democracy. Our friend Larry Lessig
talks about how we just feel so disengaged
from the process, and we’ve become increasingly
cynical about it. Just don’t do that, please. ERIK BRYNJOLFSSON: You
know, technology– go ahead. AUDIENCE: I think I
meant in between, so– ERIK BRYNJOLFSSON: Yeah, yeah. Let me try and go
to that, because I think that the technologists
and engineers have been great at solving
problems, and we talked to some of the folks that
responded to the first DARPA Grand Challenge, which
was, as you know probably, to build a driverless car,
which many people thought was impossible,
and they responded. The new DARPA Urban Challenge
is a robot– six-foot, Atlas robot– that weighs about
300 pounds that actually will be able to walk up to a
vehicle, open the door, get inside, drive it, get
out, walk across rubble, up some stairs, attach a hose,
a whole set of other tasks. And again, at first it
seemed like science-fiction impossibility, but our friends
at the MIT Computer Science and AI Lab and elsewhere seem
to be making headway on that. It may be that we need
to think about a new kind of grand challenge that isn’t to
make an amazing new technology, but is just to think
about how we can reinvent an economic, social,
organizational system– one that creates as many
jobs as it’s automating. Many of the previous
engineering tasks have either implicitly or quite
explicitly been designed to answer the
question, how can we take labor out of this process? How can we make this
more labor-saving, and have a machine do what a
person used to be able to do? But technology
doesn’t have to just be a substitute for humans. Many technologies can be
complements for humans, and can enhance the
value of humans, and maybe if we put
more focused effort into a grand challenge for
creating technologies that boost the amounts of labor,
that are complements to labor, and allow people to do things
they couldn’t have done before, then we are likely to get that
kind of shared-prosperity world that we were hoping for. And so that would be the
kind of grand challenge. We haven’t quite figured out
how to formulate that right, but maybe if we
could do that, it would inspire many
of you and people elsewhere in the
country and the world to try to solve that
problem, as opposed to waiting for, say,
Washington to take action. ANDREW MCAFEE: I think we have
time for one more question, but it has to be a really
bang-up final question. Are you ready? AUDIENCE: Sure. Maybe. So actually, that was a
really interesting answer, if under-specified so far,
that you started to give, because before you
started to give that, I would’ve started to
wonder if you had had the reaction– other people
had the reaction to the book that your proposed
prescriptions don’t seem of the same magnitude as the
problem that you’re describing. For the most part, you seem
to just be describing tweaks, or maybe if you
have one big idea, it’s education, which everybody
has talked about for years, and nobody knows how to fix,
and if you could fix it, it wouldn’t pay off for
decades down the line anyway, and you have new
problems by that point. So what do you think is the
biggest policy prescription or advice for the
future that you have that is proportionate to
the magnitude of the problem? ANDREW MCAFEE: And the nastiest
and maybe most accurate comment we’ve gotten
about our work is, you guys are
proposing linear solutions to an exponential situation. Ow, right? So we’re mindful of it. What we don’t want
to do is say that we know what the trajectory of
technology is going to be, and therefore what kind
of moonshotty intervention is appropriate to get there. What we’re saying instead is
essentially, let’s iterate. Let’s tinker. Let’s mess with the
things that we know will move the dial in
the right direction, and let’s make sure that we’re
tracking the problem correctly, as opposed to trying to
anticipate where this thing is going to be in some crazy
time frame down the road. I take a lot of insight from
your chairman’s repeated assertion that the crazy,
long-frame planning horizon for Google
is five years. Overshooting that by a lot
with our policy recommendations seems like a recipe for
failure or disaster. ERIK BRYNJOLFSSON:
I mean, you’ve raised a very
important point, which is that we don’t have a
silver-bullet solution to this. I wish we did,
but we can’t point to one silver-bullet
solution, and that partly reflects how we went
about writing this book. Many people who write this kind
of book start with a policy, and then they go
and write a bunch of reasons why this policy
needs to be put in place. ANDREW MCAFEE: Because
they’re policy guys. ERIK BRYNJOLFSSON: Right. We are not policy
guys, and we focused on diagnosing what’s going on. We want to lay those issues out. We have three chapters’ worth
of potential solutions that we think will help
with the problem, but ultimately we think that
by start this conversation, maybe you and others in this
room and elsewhere will help develop the solutions. We lay out some of
the problems there, and this is the kind of thing
that, just as in the past, we hope to crowdsource
it, in essence. And we have tried to walk
the walk a little bit, in terms of starting
a new initiative on the digital economy at MIT,
where we were bringing together people to grapple
with these questions, and we’re hoping we
can make some progress. But I have to confess,
I can’t say here, if you just implemented
the Earned Income Tax Credit, the problem
will be solved, because it’s not
going to be that easy. AUDIENCE: A very
brief follow-up– when you’re in a
ditch, stop digging. What’s the single-most
thing that we shouldn’t do to make the problem worse
than maybe we are already doing? ANDREW MCAFEE:
Turning out people who do not have the skills to be
valuable workers in the economy of today, let alone tomorrow. The business leaders that we
talk to are so frustrated. Up and down the
skill ladder, they can’t find the people
that they need, even for the positions they have
open, even here in America. That just doesn’t
make any sense. ERIK BRYNJOLFSSON:
I would second that. I would also say
that there’s been a problem of siloing
technical people, from the economists
and the policy makers. We find that these groups that
we’ve tried to bridge just don’t speak the same language. They don’t even recognize
the same problems, and if we could get people
coordinating on those, I think that would
be half the battle. ANDREW MCAFEE: Thank
you all for coming. This is fantastic. [APPLAUSE]

16 thoughts on “Erik Brynjolfsson & Andrew McAfee: “The Second Machine Age” | Talks at Google

  1. Outdated, shallow, ideological disappointment.
    Their good intentions are obvious but its still disappointing. I (as literally nobody) could have given this lecture at least 8 years ago (when most "economists" still pretended they dont know what "structural unemployment" actually is), with better information (US wages, inflation adjusted, are stagnant from ~1970) and genuine, non-ideological tackling of topics (outdated, prostitution-based culture as real obstacle for application of technology). Thats not meant to show how great I am but how mediocre and disappointing this talk was. Today, when you only need the eyes to see what is happening "experts" are still holding their heads in the sand. Hopefully, only in lectures such as this one.

  2. I think they were not telling what they really want to, that is that we can't learn that fast to keep up with how fast things are changing.  Governments are going into deep debt subsidizing industries to just keep people employed.   

  3. The solution is automation socialism: let's all get paid simply for being alive. Work is a dated concept. Let the robots toil.

  4. MEANWHILE, in the other video someone is bored because he has too much stuff and wants to tell us how this problem of having "too much stuff in your hands" is the biggest problem we face in the 21st century.

  5. This has been the best 60 min + I have spent on youtube!! Mandatory discussion!!! Great minds! Just downloaded their audio book, 8 hours of intersesting stuff waiting for me

  6. This will lead to a future of working less. We need to get out of the mind set of working hard for a living when machines are more productive than us and we have more free time. However, the transition to this is incredibly disruptive because it requires sweeping changes to our way of life. We need to tackle the issue about this transition now.

  7. So he talks about income tax credits, giving people money, then turns around and says he supports property rights and doesn't want people to be slaves…

    How will you give people money they didn't earn without enslaving the productive individuals and going against their property rights? Have the robots print more money maybe? Well that devalues money for those who earned it and is just another form of theft.

    Very dissapointed in this video. Especially when talking about how gov should handle schools. How about you get gov out of the way. Learn some things from Michael B Horn.

  8. The opening speaker has NO understanding of economics. If we were still on the gold standard (ended in 1971) we would be in great shape. When gas was $1/gal gold was $35.oz. 1 oz gold bought 35 gal. Now 1 coin wiould buy 390 gal. Gas is getting cheaper. The dollar is collapsing. The other problems are a result of that.

  9. One good solution for this "problem", predicted by Tesla, is to create a great and powerful ministry of health, not "the big pharma", but one based on physical education and sports, paying wages according.
    Everyone investing in a better health and respectively wages will lead society to a better well-being Human being.

  10. This is a very interesting and difficult topic. I actually studied at MIT Sloan for my MBA and remember Professor Brynjolfsson. I'm not as familiar with Andrew McAfee, but I'm certain he's every bit as astute and thoughtful. Framing the discussion as protecting the past from the future vs protecting the future from the past is a great way to distill the issue at hand. As I'm sure both authors are aware, however, there is no future without a past. Lobbying groups such as professional associations and other interest groups (for all their flaws) sometimes serve to keep society from breaking; and allowing artificial intelligence to rapidly supplant jobs involving high-level cognitive skills is something that could break society. The technology and dimensions of the risk are very different, but the scope of artificial intelligence's potential impact (good and bad) is somewhat analogous to that of nuclear energy. I agree that the technology should be allowed to develop, but I think a game plan needs to be in place. Otherwise, the animosity that these technologies might arouse could be incredibly destructive. Companies that develop these technologies need to acknowledge the severe disruptions that could ensue (thankfully, it seemed many of the employees at Google are well aware of the ramifications of such technologies). Hopefully, this will minimize polarization around the issue and foster constructive dialogue about how such technologies can be introduced in order to benefit society as a whole.

  11. i like the idea of inventing ways for humans and machines to work together to make new jobs for the future. lets re-invent the future of our society, agree with that. x-contests?

  12. Love this ❤️have this conversation with my high school daughter. Going into the real world as a college student . Thank you for sharing . Love the book .

Leave a Reply

Your email address will not be published. Required fields are marked *