you might recognize this robot as an
image from the movie Terminator
science-fiction has cultivated fear that
an AI will take over and ultimately lead
to her untimely demise my message is
simple robots will help humanity so for
our live audience the human robot dance
to it that we just watched is incredibly
moving
but this robot does not fall into the
section of artificial intelligence and
it is because it was programmed exactly
to have very specific movements and
those movements were carried out just as
it was programmed in this way it’s
acting more as a machine rather than an
intelligent entity so for something to
qualify as artificial intelligence it
has to be capable of learning for over
300 years the word computer was used to
describe humans carrying out
calculations and it wasn’t until about
80 years ago with modern day digital
computers that we started to use the
word computer as describing a machine
carrying out a calculation now we’re
entering the age of artificial
intelligence where machines are capable
of learning so there’s two main
subsections of artificial intelligence
one is software the other is hardware so
software is going to be line by line
code and essentially what it is is it
takes a system and allows for training
from a data set this is a learning
process
and then it goes a little bit further
and in this area it’s allowed to make
classifications or predictions depending
on what the task is hardware AI on the
other hand is a physical chip capable of
learning and it follows the same exact
process as the software does it should
definitely note that software AI is all
over the internet all the major
companies are already using it
hardware AI however is being researched
by all the major companies I have been
working on Hardware AI the last five
years and a great example of software a
AI is self-driving cars so 1.3 million
people die every year in automobile
accidents and 20 to 50 million people
are disabled or injured this is a number
one leading cause of accidental death in
the world
so self-driving cars are the future they
will never have a lapse in intention
they will never fall asleep never drive
drunk or recklessly they will virtually
remove automobile accidents as a cause
of death so we can see the software AI
is going to save lives and now the
question is can a I help save our
environment growing up in Washington
State on the coastline I have an extreme
of nature and it’s very important to me
to give back to the world that we live
in so it turns out hardware AI will lead
to robots that are able to work in real
time on site and determine say a
cigarette is a piece of trash
but leave a piece of driftwood where it
is on top of this we have plastic piling
up in the oceans and eventually we’re
going to have massive ocean bots that
will pick up plastic and leave fish
alone also we have drones but using AI
software called swarm intelligence
modeled after birds and B movement we
can quickly map out oil spills and then
deploy see drones clean these up this is
a much cleaner alternative than current
oil spill techniques so the real
question is why can’t we just use
standard computers for this and there’s
two main problems with standard
computers so number one you have a
separation of memory and processing and
this space between it causes a
limitation of information transfer
number two the elements in your phones
and computers today are so small
we can’t really make them much smaller
so we’re we’re hitting the limit of
standard computer architectures so we
need an alternative architecture we want
to combine this memory and processing
into the same platform and we want it to
be super super low power so if we look
to nature is an inspiration it turns out
the human brain does just this very low
power combining memory and processing we
are able to store memories and solve
complex problems and this is done by
electrical signals flowing through
neurons in the brain which are highly
interconnected and densely packed so my
group has sought to do just this
starting with a standard chip if we zoom
into this red box we can see platinum
electrodes and these are just here to
monitor electrical activity from here
another zoom in we put down tiny tiny
little seed sites and from these little
seeds we put them in a solution and add
on my atom these seeds are traded out
and self assembled to create an
incredibly dense interconnected Network
very similar to the human brain we get
about a billion connections per square
centimeter when we make these not only
that every single chip that we grow is
going to be incredibly different from
the last just like everyone’s brains are
different from each others but they have
similar density and similar properties
and have long range connections short
range connections and it turns out this
is a zoom in at the last photo if you
look at every place that a nano where
crosses another nano wire they can build
up atom by atom and form these tiny
filaments and this is a switch and if
the filament is very
we call that short-term memory and if
it’s very thick we call it long-term
memory so just like in the mammalian
brain and standard computers today are
more like a switch it turns on and it
turns off
there’s no range of the connection there
so I’m gonna leave you guys with an
image of actual human neurons they are
highly interconnected and you can see
they do look pretty similar to some of
the networks that I’ve grown so with
this I would like to ask you given the
potential benefits to society and the
environment please don’t be afraid to
say yes to AI
[Applause]
[Music]
[Applause]
[Music]
image from the movie Terminator
science-fiction has cultivated fear that
an AI will take over and ultimately lead
to her untimely demise my message is
simple robots will help humanity so for
our live audience the human robot dance
to it that we just watched is incredibly
moving
but this robot does not fall into the
section of artificial intelligence and
it is because it was programmed exactly
to have very specific movements and
those movements were carried out just as
it was programmed in this way it’s
acting more as a machine rather than an
intelligent entity so for something to
qualify as artificial intelligence it
has to be capable of learning for over
300 years the word computer was used to
describe humans carrying out
calculations and it wasn’t until about
80 years ago with modern day digital
computers that we started to use the
word computer as describing a machine
carrying out a calculation now we’re
entering the age of artificial
intelligence where machines are capable
of learning so there’s two main
subsections of artificial intelligence
one is software the other is hardware so
software is going to be line by line
code and essentially what it is is it
takes a system and allows for training
from a data set this is a learning
process
and then it goes a little bit further
and in this area it’s allowed to make
classifications or predictions depending
on what the task is hardware AI on the
other hand is a physical chip capable of
learning and it follows the same exact
process as the software does it should
definitely note that software AI is all
over the internet all the major
companies are already using it
hardware AI however is being researched
by all the major companies I have been
working on Hardware AI the last five
years and a great example of software a
AI is self-driving cars so 1.3 million
people die every year in automobile
accidents and 20 to 50 million people
are disabled or injured this is a number
one leading cause of accidental death in
the world
so self-driving cars are the future they
will never have a lapse in intention
they will never fall asleep never drive
drunk or recklessly they will virtually
remove automobile accidents as a cause
of death so we can see the software AI
is going to save lives and now the
question is can a I help save our
environment growing up in Washington
State on the coastline I have an extreme
of nature and it’s very important to me
to give back to the world that we live
in so it turns out hardware AI will lead
to robots that are able to work in real
time on site and determine say a
cigarette is a piece of trash
but leave a piece of driftwood where it
is on top of this we have plastic piling
up in the oceans and eventually we’re
going to have massive ocean bots that
will pick up plastic and leave fish
alone also we have drones but using AI
software called swarm intelligence
modeled after birds and B movement we
can quickly map out oil spills and then
deploy see drones clean these up this is
a much cleaner alternative than current
oil spill techniques so the real
question is why can’t we just use
standard computers for this and there’s
two main problems with standard
computers so number one you have a
separation of memory and processing and
this space between it causes a
limitation of information transfer
number two the elements in your phones
and computers today are so small
we can’t really make them much smaller
so we’re we’re hitting the limit of
standard computer architectures so we
need an alternative architecture we want
to combine this memory and processing
into the same platform and we want it to
be super super low power so if we look
to nature is an inspiration it turns out
the human brain does just this very low
power combining memory and processing we
are able to store memories and solve
complex problems and this is done by
electrical signals flowing through
neurons in the brain which are highly
interconnected and densely packed so my
group has sought to do just this
starting with a standard chip if we zoom
into this red box we can see platinum
electrodes and these are just here to
monitor electrical activity from here
another zoom in we put down tiny tiny
little seed sites and from these little
seeds we put them in a solution and add
on my atom these seeds are traded out
and self assembled to create an
incredibly dense interconnected Network
very similar to the human brain we get
about a billion connections per square
centimeter when we make these not only
that every single chip that we grow is
going to be incredibly different from
the last just like everyone’s brains are
different from each others but they have
similar density and similar properties
and have long range connections short
range connections and it turns out this
is a zoom in at the last photo if you
look at every place that a nano where
crosses another nano wire they can build
up atom by atom and form these tiny
filaments and this is a switch and if
the filament is very
we call that short-term memory and if
it’s very thick we call it long-term
memory so just like in the mammalian
brain and standard computers today are
more like a switch it turns on and it
turns off
there’s no range of the connection there
so I’m gonna leave you guys with an
image of actual human neurons they are
highly interconnected and you can see
they do look pretty similar to some of
the networks that I’ve grown so with
this I would like to ask you given the
potential benefits to society and the
environment please don’t be afraid to
say yes to AI
[Applause]
[Music]
[Applause]
[Music]
Please follow and like us: