let me take you first to 1956 so this is a workshop at the Dartmouth summer
a workshop at the Dartmouth summer
a workshop at the Dartmouth summer school in Vermont and it’s considered
school in Vermont and it’s considered
school in Vermont and it’s considered date of birth of artificial intelligence
date of birth of artificial intelligence
date of birth of artificial intelligence and let me read the statement they made
and let me read the statement they made
and let me read the statement they made why they had this meeting to proceed on
why they had this meeting to proceed on
why they had this meeting to proceed on the basis of a conjecture that every
the basis of a conjecture that every
the basis of a conjecture that every aspect of learning or any other feature
aspect of learning or any other feature
aspect of learning or any other feature of intelligence can be in principle be
of intelligence can be in principle be
of intelligence can be in principle be so precisely described that the machine
so precisely described that the machine
so precisely described that the machine can be made to simulate it think about
can be made to simulate it think about
can be made to simulate it think about it they want to crack intelligence they
it they want to crack intelligence they
it they want to crack intelligence they want to have a computer do what humans
want to have a computer do what humans
want to have a computer do what humans do
do
do that’s 1956 that’s a bold goal because
that’s 1956 that’s a bold goal because
that’s 1956 that’s a bold goal because computer in 1956 is like this hmm so
computer in 1956 is like this hmm so
computer in 1956 is like this hmm so when you have this and you think about
when you have this and you think about
when you have this and you think about making a computer simulated intelligence
making a computer simulated intelligence
making a computer simulated intelligence you really have a vision think about it
you really have a vision think about it
you really have a vision think about it if you swipe on your iPhone the
if you swipe on your iPhone the
if you swipe on your iPhone the computing power to do that simple action
computing power to do that simple action
computing power to do that simple action might take this machine months so a
might take this machine months so a
might take this machine months so a session on Facebook may take 15 years so
session on Facebook may take 15 years so
session on Facebook may take 15 years so it’s a long shot to think of artificial
it’s a long shot to think of artificial
it’s a long shot to think of artificial intelligence in that environment and in
intelligence in that environment and in
intelligence in that environment and in fact it took something like 50 to 60
fact it took something like 50 to 60
fact it took something like 50 to 60 years before that promise could be
years before that promise could be
years before that promise could be materialized so we are more or less in
materialized so we are more or less in
materialized so we are more or less in 2012 2010 2012 that the first wave of
2012 2010 2012 that the first wave of
2012 2010 2012 that the first wave of big important results came out the first
big important results came out the first
big important results came out the first wave of important algorithms that sort
wave of important algorithms that sort
wave of important algorithms that sort of simulate the brain and do something
of simulate the brain and do something
of simulate the brain and do something like the brain do does better image
like the brain do does better image
like the brain do does better image recognition is the first big example
recognition is the first big example
recognition is the first big example image recognition implies that you have
image recognition implies that you have
image recognition implies that you have an image a picture of something and a
an image a picture of something and a
an image a picture of something and a machine tells you what’s in there right
machine tells you what’s in there right
machine tells you what’s in there right so this has been worked on for many many
so this has been worked on for many many
so this has been worked on for many many years and you can see the albums were
years and you can see the albums were
years and you can see the albums were okay for a long time they could
okay for a long time they could
okay for a long time they could recognize three out of four images which
recognize three out of four images which
recognize three out of four images which is not very good for doing things this
is not very good for doing things this
is not very good for doing things this is just plain software and then a group
is just plain software and then a group
is just plain software and then a group of researchers in Toronto say forget
of researchers in Toronto say forget
of researchers in Toronto say forget about that way of doing things we don’t
about that way of doing things we don’t
about that way of doing things we don’t take an algorithm a software which you
take an algorithm a software which you
take an algorithm a software which you have put all the instruction to say how
have put all the instruction to say how
have put all the instruction to say how to recognize an image but we teach the
to recognize an image but we teach the
to recognize an image but we teach the algorithm we take an algorithm which is
algorithm we take an algorithm which is
algorithm we take an algorithm which is actually designed to learn not to do
actually designed to learn not to do
actually designed to learn not to do things but to learn what happens is that
things but to learn what happens is that
things but to learn what happens is that from 75% it goes up to 80% and then year
from 75% it goes up to 80% and then year
from 75% it goes up to 80% and then year after year it becomes super human so it
after year it becomes super human so it
after year it becomes super human so it means that does a human task better than
means that does a human task better than
means that does a human task better than a human and now you can do that in image
a human and now you can do that in image
a human and now you can do that in image recognition in text recognition
recognition in text recognition
recognition in text recognition in speech in motion and so on and so on
in speech in motion and so on and so on
in speech in motion and so on and so on and so on
and so on
and so on in the past everything we did with
in the past everything we did with
in the past everything we did with software was instructions you tell us
software was instructions you tell us
software was instructions you tell us after what to do and everything you do
after what to do and everything you do
after what to do and everything you do with the software has to be told to the
with the software has to be told to the
with the software has to be told to the software by somebody so when you click a
software by somebody so when you click a
software by somebody so when you click a key on a keyboard somebody has told the
key on a keyboard somebody has told the
key on a keyboard somebody has told the machine to do everything that happens
machine to do everything that happens
machine to do everything that happens until something comes up on the screen
until something comes up on the screen
until something comes up on the screen except for bugs right when you have bugs
except for bugs right when you have bugs
except for bugs right when you have bugs that’s the only time when the software
that’s the only time when the software
that’s the only time when the software invents something even if it’s something
invents something even if it’s something
invents something even if it’s something that you don’t really want so the new
that you don’t really want so the new
that you don’t really want so the new set of algorithms is but has been are
set of algorithms is but has been are
set of algorithms is but has been are based on a different premise the premise
based on a different premise the premise
based on a different premise the premise that you copied the way the brain works
that you copied the way the brain works
that you copied the way the brain works and the brain works in this way many
and the brain works in this way many
and the brain works in this way many simple pieces of computing which are
simple pieces of computing which are
simple pieces of computing which are core neurons which are compiled together
core neurons which are compiled together
core neurons which are compiled together in a very complex layer and this is how
in a very complex layer and this is how
in a very complex layer and this is how it happens
it happens
it happens within this image recognition algorithm
within this image recognition algorithm
within this image recognition algorithm so you put an image at the entrance that
so you put an image at the entrance that
so you put an image at the entrance that image is segments in different places
image is segments in different places
image is segments in different places and after you’ve shown many many images
and after you’ve shown many many images
and after you’ve shown many many images that algorithm can make the association
that algorithm can make the association
that algorithm can make the association between the input and the output you
between the input and the output you
between the input and the output you show that image and this is Sarah is so
show that image and this is Sarah is so
show that image and this is Sarah is so another image and this is another person
another image and this is another person
another image and this is another person Facebook does that all the time so
Facebook does that all the time so
Facebook does that all the time so whenever you post something in facebook
whenever you post something in facebook
whenever you post something in facebook the first thing that happens if there is
the first thing that happens if there is
the first thing that happens if there is a picture there it goes through a series
a picture there it goes through a series
a picture there it goes through a series of neural networks which actually split
of neural networks which actually split
of neural networks which actually split out the content because of that you have
out the content because of that you have
out the content because of that you have a change in the parting you go from
a change in the parting you go from
a change in the parting you go from software or algorithms in the old way in
software or algorithms in the old way in
software or algorithms in the old way in which you actually do instructions you
which you actually do instructions you
which you actually do instructions you tell the algorithm everything that you
tell the algorithm everything that you
tell the algorithm everything that you want it to do that’s why you call
want it to do that’s why you call
want it to do that’s why you call instructions in software you go from the
instructions in software you go from the
instructions in software you go from the earth to what we call supervised
earth to what we call supervised
earth to what we call supervised learning so you show algorithms many
learning so you show algorithms many
learning so you show algorithms many many examples of images of speech of
many examples of images of speech of
many examples of images of speech of applications for university and so on
applications for university and so on
applications for university and so on and that algorithm recognizes the
and that algorithm recognizes the
and that algorithm recognizes the relationship between something that says
relationship between something that says
relationship between something that says this and the output it makes the
this and the output it makes the
this and the output it makes the connection between input and output and
connection between input and output and
connection between input and output and then you show it something it hasn’t
then you show it something it hasn’t
then you show it something it hasn’t seen and it can make it itself it’s a
seen and it can make it itself it’s a
seen and it can make it itself it’s a very very simple way of doing
very very simple way of doing
very very simple way of doing intelligence it’s actually very very
intelligence it’s actually very very
intelligence it’s actually very very simple this is not how our brain works
simple this is not how our brain works
simple this is not how our brain works in practice we do many other things but
in practice we do many other things but
in practice we do many other things but this is what we have now so it’s not
this is what we have now so it’s not
this is what we have now so it’s not really artificial intelligence what we
really artificial intelligence what we
really artificial intelligence what we call restricted artificial intelligence
call restricted artificial intelligence
call restricted artificial intelligence is extremely powerful nonetheless where
is extremely powerful nonetheless where
is extremely powerful nonetheless where we are going to go it’s unsupervised
we are going to go it’s unsupervised
we are going to go it’s unsupervised learning
learning
learning so you need very few examples to learn
so you need very few examples to learn
so you need very few examples to learn think about the first time you saw a
think about the first time you saw a
think about the first time you saw a Segway right you just needed one frame
Segway right you just needed one frame
Segway right you just needed one frame one second of that object to figure out
one second of that object to figure out
one second of that object to figure out it was in the same category of scape
it was in the same category of scape
it was in the same category of scape works bicycles and so on you didn’t need
works bicycles and so on you didn’t need
works bicycles and so on you didn’t need anything else just one example computers
anything else just one example computers
anything else just one example computers cannot do that it might take years or
cannot do that it might take years or
cannot do that it might take years or decades for them to learn it or maybe
decades for them to learn it or maybe
decades for them to learn it or maybe never but in any case so this is the
never but in any case so this is the
never but in any case so this is the part where we are so this moment we know
part where we are so this moment we know
part where we are so this moment we know how to do supervised learning we know
how to do supervised learning we know
how to do supervised learning we know how to teach machines from many examples
how to teach machines from many examples
how to teach machines from many examples to do it better than humans and that’s
to do it better than humans and that’s
to do it better than humans and that’s powerful it’s powerful because if a
powerful it’s powerful because if a
powerful it’s powerful because if a person can do it task a mental task in
person can do it task a mental task in
person can do it task a mental task in less than a second probably it can be
less than a second probably it can be
less than a second probably it can be automated probably there’s an algorithm
automated probably there’s an algorithm
automated probably there’s an algorithm that does it better than you but not in
that does it better than you but not in
that does it better than you but not in ten years not in five years with what we
ten years not in five years with what we
ten years not in five years with what we have now technology this is now if you
have now technology this is now if you
have now technology this is now if you take that app step further
take that app step further
take that app step further what is the impact why is it important
what is the impact why is it important
what is the impact why is it important why it’s important because about 60% of
why it’s important because about 60% of
why it’s important because about 60% of jobs occupations and so on will be
jobs occupations and so on will be
jobs occupations and so on will be changed by that without any development
changed by that without any development
changed by that without any development anything new just what we have now so
anything new just what we have now so
anything new just what we have now so all my job your future jobs and so on
all my job your future jobs and so on
all my job your future jobs and so on will be affected by this in a
will be affected by this in a
will be affected by this in a substantial fundamental way so in five
substantial fundamental way so in five
substantial fundamental way so in five years from now it would be impossible to
years from now it would be impossible to
years from now it would be impossible to imagine the doctor that does diagnoses
imagine the doctor that does diagnoses
imagine the doctor that does diagnoses without consulting an algorithm that’s
without consulting an algorithm that’s
without consulting an algorithm that’s just impossible to imagine it’s just
just impossible to imagine it’s just
just impossible to imagine it’s just impossible to conceive a lawyer that
impossible to conceive a lawyer that
impossible to conceive a lawyer that starts the case without an agent doing
starts the case without an agent doing
starts the case without an agent doing the background work for the lawyer
the background work for the lawyer
the background work for the lawyer itself so it’s important so the question
itself so it’s important so the question
itself so it’s important so the question is can we trust these algorithms do they
is can we trust these algorithms do they
is can we trust these algorithms do they tell you the truth and when they make
tell you the truth and when they make
tell you the truth and when they make decisions for you do they have a moral
decisions for you do they have a moral
decisions for you do they have a moral standard that they use for making the
standard that they use for making the
standard that they use for making the decision so I’m going to touch on some
decision so I’m going to touch on some
decision so I’m going to touch on some of these elements in the presentation
of these elements in the presentation
of these elements in the presentation now the first problem you have is this
now the first problem you have is this
now the first problem you have is this many of the powerful tools that we have
many of the powerful tools that we have
many of the powerful tools that we have developed within what we call artificial
developed within what we call artificial
developed within what we call artificial intelligence are actually black boxes so
intelligence are actually black boxes so
intelligence are actually black boxes so you know that they work well you know
you know that they work well you know
you know that they work well you know what you put in you know what you put
what you put in you know what you put
what you put in you know what you put out you don’t really know why they work
out you don’t really know why they work
out you don’t really know why they work that well it is surprising but we
that well it is surprising but we
that well it is surprising but we actually don’t know the real function
actually don’t know the real function
actually don’t know the real function the real inner working of a neural
the real inner working of a neural
the real inner working of a neural network we know the wiring but why does
network we know the wiring but why does
network we know the wiring but why does it usually
it usually
it usually we actually don’t know so you have to
we actually don’t know so you have to
we actually don’t know so you have to start thinking about how do you actually
start thinking about how do you actually
start thinking about how do you actually poke inside this algorithm to make sure
poke inside this algorithm to make sure
poke inside this algorithm to make sure that they do what you want and they do
that they do what you want and they do
that they do what you want and they do it in a correct way
it in a correct way
it in a correct way so let’s start with a simple case in
so let’s start with a simple case in
so let’s start with a simple case in this case let’s say that you want to
this case let’s say that you want to
this case let’s say that you want to interpret this image and you want to
interpret this image and you want to
interpret this image and you want to have an algorithmic figure out where
have an algorithmic figure out where
have an algorithmic figure out where there’s more people that’s a simple case
there’s more people that’s a simple case
there’s more people that’s a simple case right there will be no algorithm in the
right there will be no algorithm in the
right there will be no algorithm in the world and make a different assumption
world and make a different assumption
world and make a different assumption this is Obama inauguration and this is
this is Obama inauguration and this is
this is Obama inauguration and this is Trump in aberration where there’s more
Trump in aberration where there’s more
Trump in aberration where there’s more people pretty clears on the right
people pretty clears on the right
people pretty clears on the right every machine will be able to do this is
every machine will be able to do this is
every machine will be able to do this is simple it’s explicit information
simple it’s explicit information
simple it’s explicit information actually every human being would do the
actually every human being would do the
actually every human being would do the same except perhaps 10 or 12 in
same except perhaps 10 or 12 in
same except perhaps 10 or 12 in Washington so it’s fairly clear
Washington so it’s fairly clear
Washington so it’s fairly clear situation here but then what happens
situation here but then what happens
situation here but then what happens when you have this so here you have a
when you have this so here you have a
when you have this so here you have a machine that takes an algorithm that
machine that takes an algorithm that
machine that takes an algorithm that takes many many medical dossiers
takes many many medical dossiers
takes many many medical dossiers compiles them and it’s capable of
compiles them and it’s capable of
compiles them and it’s capable of detecting sepsis which is you know a
detecting sepsis which is you know a
detecting sepsis which is you know a serious condition 90% of the time better
serious condition 90% of the time better
serious condition 90% of the time better or in actually faster than a doctor so
or in actually faster than a doctor so
or in actually faster than a doctor so you have an algorithm that actually
you have an algorithm that actually
you have an algorithm that actually performs better than a doctor in the
performs better than a doctor in the
performs better than a doctor in the sense that it detects things quicker
sense that it detects things quicker
sense that it detects things quicker right and then you doctor assume you’re
right and then you doctor assume you’re
right and then you doctor assume you’re a doctor you have to find out what do I
a doctor you have to find out what do I
a doctor you have to find out what do I do with that do I trust it do I act upon
do with that do I trust it do I act upon
do with that do I trust it do I act upon it because I got 16 hours advantage but
it because I got 16 hours advantage but
it because I got 16 hours advantage but I have to find out if this algorithm is
I have to find out if this algorithm is
I have to find out if this algorithm is correct so what is it that made this
correct so what is it that made this
correct so what is it that made this algorithm go from the evidence it had to
algorithm go from the evidence it had to
algorithm go from the evidence it had to the conclusion it makes so you have to
the conclusion it makes so you have to
the conclusion it makes so you have to have that clear because otherwise you
have that clear because otherwise you
have that clear because otherwise you can’t use it it gets even more
can’t use it it gets even more
can’t use it it gets even more complicated in situations like this so
complicated in situations like this so
complicated in situations like this so this is an algorithm that predicts the
this is an algorithm that predicts the
this is an algorithm that predicts the road accidents by the way this is a
road accidents by the way this is a
road accidents by the way this is a simulation we cannot put out in the real
simulation we cannot put out in the real
simulation we cannot put out in the real panels on the roads because we don’t
panels on the roads because we don’t
panels on the roads because we don’t know how people would react to this but
know how people would react to this but
know how people would react to this but these are algorithms anyway that
these are algorithms anyway that
these are algorithms anyway that predicted in the next 10 kilometers in
predicted in the next 10 kilometers in
predicted in the next 10 kilometers in the next 30 minute there’s going to be
the next 30 minute there’s going to be
the next 30 minute there’s going to be an accident or not right and then if
an accident or not right and then if
an accident or not right and then if they do that how should you need to be
they do that how should you need to be
they do that how should you need to be before you actually put out that
before you actually put out that
before you actually put out that information because once the information
information because once the information
information because once the information is out and maybe this is not the best
is out and maybe this is not the best
is out and maybe this is not the best way to do it that this is going to
way to do it that this is going to
way to do it that this is going to change the way people behave so you
change the way people behave so you
change the way people behave so you might have a change of traffic which we
might have a change of traffic which we
might have a change of traffic which we proved remove the incident but you know
proved remove the incident but you know
proved remove the incident but you know it might actually make it work so you
it might actually make it work so you
it might actually make it work so you don’t really know and worse in even
don’t really know and worse in even
don’t really know and worse in even worse in any case even deeper is that
worse in any case even deeper is that
worse in any case even deeper is that you have machines that are really so
you have machines that are really so
you have machines that are really so strong and powerful like this one that
strong and powerful like this one that
strong and powerful like this one that can raise a person that cannot do that
can raise a person that cannot do that
can raise a person that cannot do that but as much as they can raise
but as much as they can raise
but as much as they can raise also crush it so really have to trust
also crush it so really have to trust
also crush it so really have to trust them completely so here’s the issue that
them completely so here’s the issue that
them completely so here’s the issue that we try to solve and I’ll give you a
we try to solve and I’ll give you a
we try to solve and I’ll give you a couple of examples how we inside the
couple of examples how we inside the
couple of examples how we inside the research community practitioners those
research community practitioners those
research community practitioners those work in AI try to do it
work in AI try to do it
work in AI try to do it the main problem you have is you have
the main problem you have is you have
the main problem you have is you have two types of algorithms does that work
two types of algorithms does that work
two types of algorithms does that work extremely well but you don’t really know
extremely well but you don’t really know
extremely well but you don’t really know why those that you know why they work
why those that you know why they work
why those that you know why they work but they don’t do it that well so
but they don’t do it that well so
but they don’t do it that well so explain ability and performance are in
explain ability and performance are in
explain ability and performance are in conflict there’s a trade-off you want
conflict there’s a trade-off you want
conflict there’s a trade-off you want good performance you don’t know how to
good performance you don’t know how to
good performance you don’t know how to explain it you want to explain it we
explain it you want to explain it we
explain it you want to explain it we have to give up performance we want to
have to give up performance we want to
have to give up performance we want to get rid of that conflict we don’t want
get rid of that conflict we don’t want
get rid of that conflict we don’t want that trade-off we want to make sure that
that trade-off we want to make sure that
that trade-off we want to make sure that actually explainable and they work well
actually explainable and they work well
actually explainable and they work well so some of the things that people are
so some of the things that people are
so some of the things that people are doing is of course to test what are the
doing is of course to test what are the
doing is of course to test what are the limits of these algorithms the first
limits of these algorithms the first
limits of these algorithms the first thing you want to do since we are
thing you want to do since we are
thing you want to do since we are talking about training and so you’re not
talking about training and so you’re not
talking about training and so you’re not coding an hour you’re training an
coding an hour you’re training an
coding an hour you’re training an algorithm are you introducing biases are
algorithm are you introducing biases are
algorithm are you introducing biases are you introducing biases that make the
you introducing biases that make the
you introducing biases that make the algorithm work in a way which is not
algorithm work in a way which is not
algorithm work in a way which is not supposed to do this is called compass
supposed to do this is called compass
supposed to do this is called compass it’s a tool used by courts in the United
it’s a tool used by courts in the United
it’s a tool used by courts in the United States some courts to decide if convict
States some courts to decide if convict
States some courts to decide if convict can be released on parole or not so you
can be released on parole or not so you
can be released on parole or not so you decide in freedom or not for a person
decide in freedom or not for a person
decide in freedom or not for a person and it does it based on history and a
and it does it based on history and a
and it does it based on history and a questionnaire and puts out the score and
questionnaire and puts out the score and
questionnaire and puts out the score and what comes out is that this algorithm
what comes out is that this algorithm
what comes out is that this algorithm this has been tested by number of people
this has been tested by number of people
this has been tested by number of people who try to push the argument to the
who try to push the argument to the
who try to push the argument to the limit is that white 2d3 event I need to
limit is that white 2d3 event I need to
limit is that white 2d3 event I need to read it because you can’t whites who d3
read it because you can’t whites who d3
read it because you can’t whites who d3 offend we’re classified as low-risk
offend we’re classified as low-risk
offend we’re classified as low-risk twice as much as blacks auditory offend
twice as much as blacks auditory offend
twice as much as blacks auditory offend so that tolerance was really biased in
so that tolerance was really biased in
so that tolerance was really biased in favor of white people right why is that
favor of white people right why is that
favor of white people right why is that is it somebody did it on purpose
is it somebody did it on purpose
is it somebody did it on purpose probably not there’s nobody wants to do
probably not there’s nobody wants to do
probably not there’s nobody wants to do that on purpose just the history the
that on purpose just the history the
that on purpose just the history the cases the judges that decided in the
cases the judges that decided in the
cases the judges that decided in the past that had that bias coded in in the
past that had that bias coded in in the
past that had that bias coded in in the information used to train the algorithm
information used to train the algorithm
information used to train the algorithm and that’s now is part of the code to
and that’s now is part of the code to
and that’s now is part of the code to actually advise judges so this is one of
actually advise judges so this is one of
actually advise judges so this is one of the main things we cannot afford to do
the main things we cannot afford to do
the main things we cannot afford to do this has to be sifted out we can do this
this has to be sifted out we can do this
this has to be sifted out we can do this because it’s it’s a matter of fact that
because it’s it’s a matter of fact that
because it’s it’s a matter of fact that worse is not a fair place it’s in just
worse is not a fair place it’s in just
worse is not a fair place it’s in just and so on but there’s one thing we
and so on but there’s one thing we
and so on but there’s one thing we cannot afford to take that and just
cannot afford to take that and just
cannot afford to take that and just encode it and make it part of algorithms
encode it and make it part of algorithms
encode it and make it part of algorithms so
so
so people are doing something experiments
people are doing something experiments
people are doing something experiments like this stretching the algorithms to
like this stretching the algorithms to
like this stretching the algorithms to the limit to find out where they fail
the limit to find out where they fail
the limit to find out where they fail where they break or in any case where
where they break or in any case where
where they break or in any case where they are biased another case of that
they are biased another case of that
they are biased another case of that exercise is this so a lot of image
exercise is this so a lot of image
exercise is this so a lot of image recognition perhaps everything that has
recognition perhaps everything that has
recognition perhaps everything that has to do with image recognition at this
to do with image recognition at this
to do with image recognition at this point in time is automatic so you just
point in time is automatic so you just
point in time is automatic so you just have software tool and Algrim that does
have software tool and Algrim that does
have software tool and Algrim that does it for you
it for you
it for you so if you have a face you recognize that
so if you have a face you recognize that
so if you have a face you recognize that that face is my face is the same with or
that face is my face is the same with or
that face is my face is the same with or without it doesn’t matter it’s still me
without it doesn’t matter it’s still me
without it doesn’t matter it’s still me right and if I change color of glasses
right and if I change color of glasses
right and if I change color of glasses it’s still me but is it the same for a
it’s still me but is it the same for a
it’s still me but is it the same for a machine actually turns out is not so
machine actually turns out is not so
machine actually turns out is not so some people are using tools and
some people are using tools and
some people are using tools and techniques to see where the algorithms
techniques to see where the algorithms
techniques to see where the algorithms break so you want to stretch it to the
break so you want to stretch it to the
break so you want to stretch it to the limit and find out places and training
limit and find out places and training
limit and find out places and training sets that fool them this is what these
sets that fool them this is what these
sets that fool them this is what these guys have found it is sufficient in this
guys have found it is sufficient in this
guys have found it is sufficient in this case to wear glasses that have a very
case to wear glasses that have a very
case to wear glasses that have a very specific color and pattern to fool the
specific color and pattern to fool the
specific color and pattern to fool the image recognition algorithm so what
image recognition algorithm so what
image recognition algorithm so what happens is that this guy on the top
happens is that this guy on the top
happens is that this guy on the top right that’s the guy but is recognized
right that’s the guy but is recognized
right that’s the guy but is recognized as the lady the middle lady is
as the lady the middle lady is
as the lady the middle lady is recognized as that guy and so on so the
recognized as that guy and so on so the
recognized as that guy and so on so the algorithm is completely confused and and
algorithm is completely confused and and
algorithm is completely confused and and the reason why it does is because that
the reason why it does is because that
the reason why it does is because that piece of information that patterns
piece of information that patterns
piece of information that patterns actually sifts through the neural
actually sifts through the neural
actually sifts through the neural network at that some layers of the
network at that some layers of the
network at that some layers of the neural network alters something
neural network alters something
neural network alters something fundamental there remember that we don’t
fundamental there remember that we don’t
fundamental there remember that we don’t really know why this works so the only
really know why this works so the only
really know why this works so the only thing that we can do is to test it to
thing that we can do is to test it to
thing that we can do is to test it to the limit so we know where it breaks it
the limit so we know where it breaks it
the limit so we know where it breaks it will never use something that you don’t
will never use something that you don’t
will never use something that you don’t know where it breaks so this is fairly
know where it breaks so this is fairly
know where it breaks so this is fairly important piece of training but it’s all
important piece of training but it’s all
important piece of training but it’s all about training so there is some limit to
about training so there is some limit to
about training so there is some limit to what you can do by testing you have to
what you can do by testing you have to
what you can do by testing you have to do something upfront
do something upfront
do something upfront this is a typical dilemma of autonomous
this is a typical dilemma of autonomous
this is a typical dilemma of autonomous car this is a vehicle that cannot stop
car this is a vehicle that cannot stop
car this is a vehicle that cannot stop anymore and it has three terrible
anymore and it has three terrible
anymore and it has three terrible choices in the next few seconds either
choices in the next few seconds either
choices in the next few seconds either you hit the person right or you hit the
you hit the person right or you hit the
you hit the person right or you hit the tree on the zebra stripes or you go
tree on the zebra stripes or you go
tree on the zebra stripes or you go against that wall and you risk the life
against that wall and you risk the life
against that wall and you risk the life of the passengers of the car which one
of the passengers of the car which one
of the passengers of the car which one you cannot stop the car this is going to
you cannot stop the car this is going to
you cannot stop the car this is going to happen in action unfortunate has
happen in action unfortunate has
happen in action unfortunate has happened to people before it’s it is one
happened to people before it’s it is one
happened to people before it’s it is one of the situations in which the outcome
of the situations in which the outcome
of the situations in which the outcome of your choice is Nega
of your choice is Nega
of your choice is Nega there’s nothing positive that Alka right
there’s nothing positive that Alka right
there’s nothing positive that Alka right so how do you choose here well if you do
so how do you choose here well if you do
so how do you choose here well if you do training the way we do now with
training the way we do now with
training the way we do now with supervised training you probably do this
supervised training you probably do this
supervised training you probably do this so this is the way you train an
so this is the way you train an
so this is the way you train an algorithm to beat a go player this is go
algorithm to beat a go player this is go
algorithm to beat a go player this is go it’s a board games much more complex
it’s a board games much more complex
it’s a board games much more complex than chess and it was told and nobody no
than chess and it was told and nobody no
than chess and it was told and nobody no algorithm could actually win a human
algorithm could actually win a human
algorithm could actually win a human player but it has been demonstrated last
player but it has been demonstrated last
player but it has been demonstrated last years now it’s a done deal how do you
years now it’s a done deal how do you
years now it’s a done deal how do you train this you take many many games play
train this you take many many games play
train this you take many many games play in the past you have the algorithm to
in the past you have the algorithm to
in the past you have the algorithm to recognize the patterns of this game and
recognize the patterns of this game and
recognize the patterns of this game and at the end this algorithm does it as the
at the end this algorithm does it as the
at the end this algorithm does it as the good players right and then actually the
good players right and then actually the
good players right and then actually the trick is we make the algorithm play
trick is we make the algorithm play
trick is we make the algorithm play against itself so it’s going to train
against itself so it’s going to train
against itself so it’s going to train even better after a million games so you
even better after a million games so you
even better after a million games so you can train a machine a car this way
can train a machine a car this way
can train a machine a car this way because then you have to create a
because then you have to create a
because then you have to create a million debts before figuring out that
million debts before figuring out that
million debts before figuring out that that’s not really what you want from the
that’s not really what you want from the
that’s not really what you want from the car so you have to find another way to
car so you have to find another way to
car so you have to find another way to do it so the way the tech tree is
do it so the way the tech tree is
do it so the way the tech tree is developing now is this you actually
developing now is this you actually
developing now is this you actually raise algorithms the way the same ways
raise algorithms the way the same ways
raise algorithms the way the same ways you raise children exactly the same way
you raise children exactly the same way
you raise children exactly the same way with one catch you cannot wait 20 years
with one catch you cannot wait 20 years
with one catch you cannot wait 20 years so you have to find ways to accelerate
so you have to find ways to accelerate
so you have to find ways to accelerate that training so there are a couple of
that training so there are a couple of
that training so there are a couple of ways that actually work well one that I
ways that actually work well one that I
ways that actually work well one that I like a lot is Georgia Tech in the u.s.
like a lot is Georgia Tech in the u.s.
like a lot is Georgia Tech in the u.s. is that with the create stories about
is that with the create stories about
is that with the create stories about the behavior of people in that situation
the behavior of people in that situation
the behavior of people in that situation thousands of stories it just narratives
thousands of stories it just narratives
thousands of stories it just narratives and you have algorithms learned from
and you have algorithms learned from
and you have algorithms learned from these narratives what are the code the
these narratives what are the code the
these narratives what are the code the moral code or the choices that people
moral code or the choices that people
moral code or the choices that people have made in those situations and then
have made in those situations and then
have made in those situations and then that way you actually have the machine
that way you actually have the machine
that way you actually have the machine learn what humans have done in a
learn what humans have done in a
learn what humans have done in a terrible situation like that it might
terrible situation like that it might
terrible situation like that it might not be the best choice but at least is
not be the best choice but at least is
not be the best choice but at least is as good as humans have done it and from
as good as humans have done it and from
as good as humans have done it and from there you can construct so there are a
there you can construct so there are a
there you can construct so there are a few experiments going on at this moment
few experiments going on at this moment
few experiments going on at this moment one that I like a lot is an algorithm
one that I like a lot is an algorithm
one that I like a lot is an algorithm like this which has been trained to
like this which has been trained to
like this which has been trained to recognize what is the choice is the
recognize what is the choice is the
recognize what is the choice is the codes that you use when you go out on a
codes that you use when you go out on a
codes that you use when you go out on a date and you go to the movies so
date and you go to the movies so
date and you go to the movies so thousands of people have constructed
thousands of people have constructed
thousands of people have constructed narratives about that event and guess
narratives about that event and guess
narratives about that event and guess what the algorithm detects a number of
what the algorithm detects a number of
what the algorithm detects a number of things but what also detects is that a
things but what also detects is that a
things but what also detects is that a key element in that schema is this so
key element in that schema is this so
key element in that schema is this so apparently for that algorithm kissing is
apparently for that algorithm kissing is
apparently for that algorithm kissing is an extremely important element of going
an extremely important element of going
an extremely important element of going out in a day to the movies based on the
out in a day to the movies based on the
out in a day to the movies based on the stories
stories
stories of many people this is a way of learning
of many people this is a way of learning
of many people this is a way of learning you know what if an algorithm turns out
you know what if an algorithm turns out
you know what if an algorithm turns out and suggested kissing I think we’re in a
and suggested kissing I think we’re in a
and suggested kissing I think we’re in a good place
good place
good place thank you
thank you
thank you [Applause]
”
Be First to Comment