imagine machines that could help trauma
and stroke victims recover and walk
again or act as our personal home
imagine machines that look like humans
or maybe look more like our four-legged
friends they could maybe help you out
around the home do your dishes for you
fill that dishwasher for you tidy up for
you maybe even do your recycling for you
all while you sit back and relax in your
home because these machines don’t need
looking after they can look after
themselves they can help you out in any
any room in your home on any flow stairs
are no problem to these machines imagine
a machine that could help set up a home
for humans on another planet this is
Valkyrie she has not been designed for
this world
she has been designed to try and help
set up a home for humans on Mars now
she’s got a lot to learn at the moment
she’s just like a toddler so there are
literally a few tumbles along the way
but she will learn she will hopefully go
to the International Space Station and
work with humans and learn from those
humans and she learned very very quickly
another great example of machine
learning because these ones have to be
trained and once they’re trained they’re
able to answer both complex problems and
create solutions to both programmed and
unprogrammed questions so another great
example of this is IBM’s Watson for on
ecology now Watson is the name of IBM’s
machine learning engine now Watson was
trained by leading oncologists from some
of the world leading cancer hospitals
along with every textbook and scientific
paper written on the subject to some 25
million in total
it’s capable of reading both structured
and unstructured data so things like
trial databases as well as doctors
handwritten notes and around 8,000
scientific journals every single day in
a study of a thousand cancer patients
Watson was able to recommend the same
treatment as the doctors in 99% of the
cases however in 30 percent of the cases
Watson was able to recommend something
better based on the scientific papers
that the human counterparts hadn’t yet
had time to read these are just some
great examples of machines and humans
working together helping each other to
try and solve some of the grand
challenges that we all face there are 14
world engineering Grand Challenges and
these include things like enhancing
virtual reality and reverse engineering
the brain engineering better medicines
providing access to clean water for
everyone and the one that I work on I
work on engineering the tools for
scientific discovery and I use radio
waves and some very smart machines to do
this one of those smart machines is
something called the Square Kilometre
Array or the SKA now the SKA is a radio
telescope that’s designed to work in
western australia and in some remote
areas in South Africa now the sheer size
of the SKA will make it 50 times more
sensitive than any other radio
instrument in the world ever and that
will allow us to answer key and
ambitious questions in astrophysics
cosmology and fundamental physics in
fact the SKA will be so sensitive it
will be able to detect an airport radar
like signal on a planet 10 light-years
away and the amount of data this thing
will generate is enormous
it’s estimated that these dishes alone
will generate more than 10 times the
global Internet traffic 10 times the
global Internet traffic if you could
take the SKA data taken in one day and
download it as a song and play that song
that song would last
two million years the high-performance
central computer will need to perform 10
to the 15 calculations every single
second images of one terabyte will need
to be transferred around the world every
single minute this indeed is a smart
machine measuring the universe and
generating huge amounts of data and data
is such an important commodity right now
it’s being mined like coal it’s being
refined like crude oil it’s being stored
in high-security vaults like gold and
it’s been sold like rolled steel but
what is data well at the fundamental
level all information is digital
consisting of a series of ones and zeros
but the information content of any any
physical system in Bates must be limited
and that’s known as something called a
bekenstein bound so in other words if
the system’s mass energy and dimensions
are known the bekenstein bound will
allow us to calculate the number of bits
required to recreate any system
perfectly right down to the quantum
level so if that’s true can we represent
the total information content of the
brain the human brain in bits it’s a
physical system after all the total
information content of the human brain
in Bates is a very big number it’s
around 2.6 times 10 to the 42 bits in
other words if you could take the
information contents of the brain and
downloaded as a song and play that song
that song would last for 50,000 trillion
trillion years what’s the most complex
object in the universe the brain imagine
machines that could think like humans I
don’t just mean from a logical
perspective but ones that were capable
of emotion and empathy these have to be
able to learn for themselves Valkyrie
the humanoid robot going to Mars she has
to learn for herself
and she has got a lot to learn but she
will learn and she learned very quickly
much more quickly than humans and that’s
what machines are very very good at deep
minds alphago is a great example of
machine learning alphago is the name of
deep Minds machine learning engine and
alphago has been designed to play the
game go now the game go is thought to
have originated in ancient China some
two and a half thousand years ago and so
it’s thought to be the oldest board game
still played today
now despite its relatively simple rules
go is actually very very complicated
game to play much more complex than
chess and alphago’s deepmind alphago’s
latest machine learning engine is a very
very smart machine indeed alphago zero
was able to learn the game go to an
expert level in just 40 days by removing
human data by removing human expertise
in any fashion the creator’s were able
to remove the constraints of human
knowledge
alphago 0 creates knowledge for itself
from first principles from a blank slate
and that allows it to be much more
powerful than any other machine before
it now despite how impressive that is
and it is extremely impressive
alphago zero can currently only do one
thing
it can play go karting play chess but in
the coming decades these machines are
getting smarter and smarter and we will
play a bigger and bigger role in our
society and they will fail along the way
but that’s ok because we as humans fail
along the way too and not only is it ok
that humans fail along the way we should
fail along the way we must fail along
the way because we have to get out of
our comfort zones and the reason we have
to get out of our comfort zones is
that’s where the magic happens you have
to fail you must fail in everything you
do please
fail it’s a great thing and celebrate
those failures they are fantastic and
don’t let anyone tell you otherwise I
fail all the time in my research and
hopefully that’s not because I’m a bad
researcher it’s because I’m trying to
push the boundaries I’m trying to
innovate in what I’m doing and in order
to innovate
I must fail I must get out of my comfort
zones
I must fail and learn from those
failures we should all learn from those
failures if we are to solve some of the
biggest challenges that we all face we
must get out of our comfort zones and we
must fail and learn from those failures
so what’s next for machines and humans
machines that look like humans
machines and humans on another planet
robots that think like us robots that
take our jobs machines that could love
the way I love my daughter what does the
future hold who knows well actually no
that’s not a statement that’s a question
who knows and the answer is you you know
you might not know it yet but it’s in
there it’s in the most complex object in
the universe and it’s just waiting for
you to fulfill your potential to
discover and create that next human
machine collaboration that changes the world thank you very much
and stroke victims recover and walk
again or act as our personal home
imagine machines that look like humans
or maybe look more like our four-legged
friends they could maybe help you out
around the home do your dishes for you
fill that dishwasher for you tidy up for
you maybe even do your recycling for you
all while you sit back and relax in your
home because these machines don’t need
looking after they can look after
themselves they can help you out in any
any room in your home on any flow stairs
are no problem to these machines imagine
a machine that could help set up a home
for humans on another planet this is
Valkyrie she has not been designed for
this world
she has been designed to try and help
set up a home for humans on Mars now
she’s got a lot to learn at the moment
she’s just like a toddler so there are
literally a few tumbles along the way
but she will learn she will hopefully go
to the International Space Station and
work with humans and learn from those
humans and she learned very very quickly
another great example of machine
learning because these ones have to be
trained and once they’re trained they’re
able to answer both complex problems and
create solutions to both programmed and
unprogrammed questions so another great
example of this is IBM’s Watson for on
ecology now Watson is the name of IBM’s
machine learning engine now Watson was
trained by leading oncologists from some
of the world leading cancer hospitals
along with every textbook and scientific
paper written on the subject to some 25
million in total
it’s capable of reading both structured
and unstructured data so things like
trial databases as well as doctors
handwritten notes and around 8,000
scientific journals every single day in
a study of a thousand cancer patients
Watson was able to recommend the same
treatment as the doctors in 99% of the
cases however in 30 percent of the cases
Watson was able to recommend something
better based on the scientific papers
that the human counterparts hadn’t yet
had time to read these are just some
great examples of machines and humans
working together helping each other to
try and solve some of the grand
challenges that we all face there are 14
world engineering Grand Challenges and
these include things like enhancing
virtual reality and reverse engineering
the brain engineering better medicines
providing access to clean water for
everyone and the one that I work on I
work on engineering the tools for
scientific discovery and I use radio
waves and some very smart machines to do
this one of those smart machines is
something called the Square Kilometre
Array or the SKA now the SKA is a radio
telescope that’s designed to work in
western australia and in some remote
areas in South Africa now the sheer size
of the SKA will make it 50 times more
sensitive than any other radio
instrument in the world ever and that
will allow us to answer key and
ambitious questions in astrophysics
cosmology and fundamental physics in
fact the SKA will be so sensitive it
will be able to detect an airport radar
like signal on a planet 10 light-years
away and the amount of data this thing
will generate is enormous
it’s estimated that these dishes alone
will generate more than 10 times the
global Internet traffic 10 times the
global Internet traffic if you could
take the SKA data taken in one day and
download it as a song and play that song
that song would last
two million years the high-performance
central computer will need to perform 10
to the 15 calculations every single
second images of one terabyte will need
to be transferred around the world every
single minute this indeed is a smart
machine measuring the universe and
generating huge amounts of data and data
is such an important commodity right now
it’s being mined like coal it’s being
refined like crude oil it’s being stored
in high-security vaults like gold and
it’s been sold like rolled steel but
what is data well at the fundamental
level all information is digital
consisting of a series of ones and zeros
but the information content of any any
physical system in Bates must be limited
and that’s known as something called a
bekenstein bound so in other words if
the system’s mass energy and dimensions
are known the bekenstein bound will
allow us to calculate the number of bits
required to recreate any system
perfectly right down to the quantum
level so if that’s true can we represent
the total information content of the
brain the human brain in bits it’s a
physical system after all the total
information content of the human brain
in Bates is a very big number it’s
around 2.6 times 10 to the 42 bits in
other words if you could take the
information contents of the brain and
downloaded as a song and play that song
that song would last for 50,000 trillion
trillion years what’s the most complex
object in the universe the brain imagine
machines that could think like humans I
don’t just mean from a logical
perspective but ones that were capable
of emotion and empathy these have to be
able to learn for themselves Valkyrie
the humanoid robot going to Mars she has
to learn for herself
and she has got a lot to learn but she
will learn and she learned very quickly
much more quickly than humans and that’s
what machines are very very good at deep
minds alphago is a great example of
machine learning alphago is the name of
deep Minds machine learning engine and
alphago has been designed to play the
game go now the game go is thought to
have originated in ancient China some
two and a half thousand years ago and so
it’s thought to be the oldest board game
still played today
now despite its relatively simple rules
go is actually very very complicated
game to play much more complex than
chess and alphago’s deepmind alphago’s
latest machine learning engine is a very
very smart machine indeed alphago zero
was able to learn the game go to an
expert level in just 40 days by removing
human data by removing human expertise
in any fashion the creator’s were able
to remove the constraints of human
knowledge
alphago 0 creates knowledge for itself
from first principles from a blank slate
and that allows it to be much more
powerful than any other machine before
it now despite how impressive that is
and it is extremely impressive
alphago zero can currently only do one
thing
it can play go karting play chess but in
the coming decades these machines are
getting smarter and smarter and we will
play a bigger and bigger role in our
society and they will fail along the way
but that’s ok because we as humans fail
along the way too and not only is it ok
that humans fail along the way we should
fail along the way we must fail along
the way because we have to get out of
our comfort zones and the reason we have
to get out of our comfort zones is
that’s where the magic happens you have
to fail you must fail in everything you
do please
fail it’s a great thing and celebrate
those failures they are fantastic and
don’t let anyone tell you otherwise I
fail all the time in my research and
hopefully that’s not because I’m a bad
researcher it’s because I’m trying to
push the boundaries I’m trying to
innovate in what I’m doing and in order
to innovate
I must fail I must get out of my comfort
zones
I must fail and learn from those
failures we should all learn from those
failures if we are to solve some of the
biggest challenges that we all face we
must get out of our comfort zones and we
must fail and learn from those failures
so what’s next for machines and humans
machines that look like humans
machines and humans on another planet
robots that think like us robots that
take our jobs machines that could love
the way I love my daughter what does the
future hold who knows well actually no
that’s not a statement that’s a question
who knows and the answer is you you know
you might not know it yet but it’s in
there it’s in the most complex object in
the universe and it’s just waiting for
you to fulfill your potential to
discover and create that next human
machine collaboration that changes the world thank you very much
Please follow and like us: