
recently I received a text message from
my sprinkler system
I repeat from my home sprinkler system
apparently it had been in communication
with the National Oceanographic and
Atmospheric Administration with regard
to the weather forecast for an area
around my home this got me to thinking
about artificial intelligence for most
of us or many of us the idea behind
artificial intelligence was suggested to
us by the 1968 film 2001 a Space Odyssey
one of the characters was called Hal and
he was a super intelligent computer who
was also in charge of the entire
spacecraft so everything was connected
to him so he was interacting with the
crew through natural language processing
in 1968 we didn’t have the technology to
build anything that could do that but
the pieces are beginning to fall into
place artificial intelligence can be
defined as behavior from a machine which
from a human being would require
intelligence so what we’re looking at is
the process of training machines to
behave in ways that appear intelligence
to us so there are three kinds of
artificial intelligence we can talk
about the first is narrow intelligence
which is what we see now self-driving
cars or the Siri interface on the iPhone
the bits and pieces to build something
like how is starting to appear in
reality the next step would be strong
artificial intelligence strong AI is
defined as a machine that is generally
as intelligent as a human being so not
only could it drive your car it could
also make pancakes for you in the
morning
it could also do your taxes so a very
general machine capable of doing a lot
of different intelligent tasks for you
and lastly there’s super strong AI under
super strong AI we began to see machines
that are improving themselves that are
redesigning their hardware and software
and
beginning to present to you solutions to
problems that you have never brought
through the machine so one of the things
that occurred to me is that AI is no
longer in the future it’s here much of
the infrastructure that we need to build
AI machines is starting to appear in our
commercial applications but specifically
the Internet the Internet is a global
communications network that forms the
fabric that underlies tying all of these
machines together the internet was
designed by the Defense Department to be
very robust so that you could attack
bits and pieces of the Internet but you
can never turn it off
the other side of that is the internet
doesn’t have an off switch now if you
watch as many science-fiction movies as
I do you have reason for concern when
you start creating intelligent machines
that do not have an off switch
the latest thing that we’ve seen is the
smart phone this is a machine that
didn’t exist a few years ago but now
there are more than a billion people who
are becoming dependent on their iPhone
or their smartphone some people are
never more than four feet away from
their phone all day so I’m thinking
about what are the impacts of this if we
think about the economic impacts they
are worthy of thought and contemplation
if we think about the transportation
industry there are over three million
people who work in the transportation
industry in the United States alone in
less than 10 years the insurance
companies are going to realize that it
is cheaper to insure a self-driving car
than to insure a car that’s driven by a
human being if we look further we think
not only is it going to be the
blue-collar world it’s also going to
impact the white-collar world IBM
already has a machine called Watson
whose job is to read papers on cancer
research so it’s looking at research
papers globally in all sorts of
different languages and it’s used as a
consultant so a doctor will type in the
the presentation of a patient’s symptoms
and it will consult with them on the
proper diagnosis and possible treatment
plans so Watson can give your doctor
advice based on a paper published
yesterday in a
no human being will ever be able to keep
up with that type of system and right
now that’s still defined as a narrow
artificial intelligence system so going
from medical it’s even going to hit the
creative arena we already have software
that writes music music but you’ve heard
if you listen to commercials and you
listen to documentaries very often some
of that music has been written by
machines recently someone completed a
motion picture and they decided to use
an artificially intelligent machine to
make the trailer for the movie so they
showed the machine a hundred trailers
from other movies and then they showed
the machine the entire motion picture
and it created the trailer that’s
currently running on TV and on the
internet to advertise a movie called
Morgan so all aspects of our economy are
going to be impacted one of my concerns
is that these systems do not fail
gracefully if we look at training versus
coding the old days and you’ve got in
front of your machine and you typed in
and you told the computer exactly what
to do today you train your machine
meaning if you want to teach a machine
to make pancakes you let it watch a
hundred people make pancakes and it will
figure out how to do that the problem
with training is that you’re never quite
sure what you’ve trained the machine to
do so even though it performs at 99.9
percent accuracy you never know when
it’s going to make a mistake and do
something embarrassing or
counterintuitive to your to your
intentions so this creates the challenge
of training how do you train a machine
to accurately do what you want and not
do things you don’t want a few years ago
a Tesla Model S made a left turn through
an 18-wheel truck sadly there was a
fatality involved the car mistook the
side of the truck for the sky because at
that moment in time they happen to be
the same color so the problem I’m
calling attention to is the fact that
it’s hard for an engineer to sit in his
or her office and imagine all possible
situations that would be clear to a
human being
but very confusing to an artificial
intelligence so what can happen beyond
that we’re thinking about a situation we
called the singularity the singularity
will occur when we have an artificial
intelligent machine that is producing
improvements to itself that no human
being knows about or understands at that
point we become vulnerable you can
imagine they’re going to be impacts and
one of them is actually very good it’s
called utopia imagine a world where no
human being ever has to do anything for
work that they’re not interested in
imagine a world where technology and
medical benefits accrue to you for free
the machines are doing it all that’s the
Utopia part but now imagine a system
failure let’s imagine the power grid
goes down the traffic lights stopped
working the telephone system stops
working if you imagine you’ve been in
this utopian state for 25 years there
may not be a human being alive who has
any idea why the power just went off and
that’s a risk that we take as we know
forward that I think we need to give
some contemplation to artificial
intelligence is going to redesign our
world it’s going to have economic
impacts and it’s going to have effects
on the human condition that we currently
cannot predict did I mention there’s no
off switch so I’m looking at we as a
society have to think about what
safeguards do we need to put in place to
make sure that this technology is
tremendously beneficial to us without
creating the risk that may be profoundly
it’s my sprinkler system he’s all
excited so I think I’m gonna have to
take this thank you for your attention [Applause]
my sprinkler system
I repeat from my home sprinkler system
apparently it had been in communication
with the National Oceanographic and
Atmospheric Administration with regard
to the weather forecast for an area
around my home this got me to thinking
about artificial intelligence for most
of us or many of us the idea behind
artificial intelligence was suggested to
us by the 1968 film 2001 a Space Odyssey
one of the characters was called Hal and
he was a super intelligent computer who
was also in charge of the entire
spacecraft so everything was connected
to him so he was interacting with the
crew through natural language processing
in 1968 we didn’t have the technology to
build anything that could do that but
the pieces are beginning to fall into
place artificial intelligence can be
defined as behavior from a machine which
from a human being would require
intelligence so what we’re looking at is
the process of training machines to
behave in ways that appear intelligence
to us so there are three kinds of
artificial intelligence we can talk
about the first is narrow intelligence
which is what we see now self-driving
cars or the Siri interface on the iPhone
the bits and pieces to build something
like how is starting to appear in
reality the next step would be strong
artificial intelligence strong AI is
defined as a machine that is generally
as intelligent as a human being so not
only could it drive your car it could
also make pancakes for you in the
morning
it could also do your taxes so a very
general machine capable of doing a lot
of different intelligent tasks for you
and lastly there’s super strong AI under
super strong AI we began to see machines
that are improving themselves that are
redesigning their hardware and software
and
beginning to present to you solutions to
problems that you have never brought
through the machine so one of the things
that occurred to me is that AI is no
longer in the future it’s here much of
the infrastructure that we need to build
AI machines is starting to appear in our
commercial applications but specifically
the Internet the Internet is a global
communications network that forms the
fabric that underlies tying all of these
machines together the internet was
designed by the Defense Department to be
very robust so that you could attack
bits and pieces of the Internet but you
can never turn it off
the other side of that is the internet
doesn’t have an off switch now if you
watch as many science-fiction movies as
I do you have reason for concern when
you start creating intelligent machines
that do not have an off switch
the latest thing that we’ve seen is the
smart phone this is a machine that
didn’t exist a few years ago but now
there are more than a billion people who
are becoming dependent on their iPhone
or their smartphone some people are
never more than four feet away from
their phone all day so I’m thinking
about what are the impacts of this if we
think about the economic impacts they
are worthy of thought and contemplation
if we think about the transportation
industry there are over three million
people who work in the transportation
industry in the United States alone in
less than 10 years the insurance
companies are going to realize that it
is cheaper to insure a self-driving car
than to insure a car that’s driven by a
human being if we look further we think
not only is it going to be the
blue-collar world it’s also going to
impact the white-collar world IBM
already has a machine called Watson
whose job is to read papers on cancer
research so it’s looking at research
papers globally in all sorts of
different languages and it’s used as a
consultant so a doctor will type in the
the presentation of a patient’s symptoms
and it will consult with them on the
proper diagnosis and possible treatment
plans so Watson can give your doctor
advice based on a paper published
yesterday in a
no human being will ever be able to keep
up with that type of system and right
now that’s still defined as a narrow
artificial intelligence system so going
from medical it’s even going to hit the
creative arena we already have software
that writes music music but you’ve heard
if you listen to commercials and you
listen to documentaries very often some
of that music has been written by
machines recently someone completed a
motion picture and they decided to use
an artificially intelligent machine to
make the trailer for the movie so they
showed the machine a hundred trailers
from other movies and then they showed
the machine the entire motion picture
and it created the trailer that’s
currently running on TV and on the
internet to advertise a movie called
Morgan so all aspects of our economy are
going to be impacted one of my concerns
is that these systems do not fail
gracefully if we look at training versus
coding the old days and you’ve got in
front of your machine and you typed in
and you told the computer exactly what
to do today you train your machine
meaning if you want to teach a machine
to make pancakes you let it watch a
hundred people make pancakes and it will
figure out how to do that the problem
with training is that you’re never quite
sure what you’ve trained the machine to
do so even though it performs at 99.9
percent accuracy you never know when
it’s going to make a mistake and do
something embarrassing or
counterintuitive to your to your
intentions so this creates the challenge
of training how do you train a machine
to accurately do what you want and not
do things you don’t want a few years ago
a Tesla Model S made a left turn through
an 18-wheel truck sadly there was a
fatality involved the car mistook the
side of the truck for the sky because at
that moment in time they happen to be
the same color so the problem I’m
calling attention to is the fact that
it’s hard for an engineer to sit in his
or her office and imagine all possible
situations that would be clear to a
human being
but very confusing to an artificial
intelligence so what can happen beyond
that we’re thinking about a situation we
called the singularity the singularity
will occur when we have an artificial
intelligent machine that is producing
improvements to itself that no human
being knows about or understands at that
point we become vulnerable you can
imagine they’re going to be impacts and
one of them is actually very good it’s
called utopia imagine a world where no
human being ever has to do anything for
work that they’re not interested in
imagine a world where technology and
medical benefits accrue to you for free
the machines are doing it all that’s the
Utopia part but now imagine a system
failure let’s imagine the power grid
goes down the traffic lights stopped
working the telephone system stops
working if you imagine you’ve been in
this utopian state for 25 years there
may not be a human being alive who has
any idea why the power just went off and
that’s a risk that we take as we know
forward that I think we need to give
some contemplation to artificial
intelligence is going to redesign our
world it’s going to have economic
impacts and it’s going to have effects
on the human condition that we currently
cannot predict did I mention there’s no
off switch so I’m looking at we as a
society have to think about what
safeguards do we need to put in place to
make sure that this technology is
tremendously beneficial to us without
creating the risk that may be profoundly
it’s my sprinkler system he’s all
excited so I think I’m gonna have to
take this thank you for your attention [Applause]
Please follow and like us: