[Music]
hey everybody how’s it going since the
dawn of human history people have had
this fear of unleashing something that
we can’t control for the ancient Greeks
this fear was manifest in the myth of
Pandora 200 years ago this year with the
publication of Frankenstein the modern
Prometheus Mary Shelley took that fear
one step further it wasn’t just about
the idea that we would unleash something
that we couldn’t control but in fact
that we might create it it’s interesting
that Shelley chose to reference
Prometheus in the subtitle of the
original book because what we think of
most often when we think of Prometheus
is the fact that he dared to steal fire
from the gods
in fact he’s described online when I was
doing my research as something of a
trickster figure which is quite
interesting because the deeper truth of
the Prometheus myth is not just that he
stole fire from the gods to give it to
man because he felt like it or he
thought that it would be fun because
Prometheus actually helped to create man
and was almost primarily responsible for
creating men this is an image of
Prometheus carving man from clay so
Prometheus stole fire from the gods to
give it to man because he felt a
responsibility to take care of the lives
that he had created in her choice to
reference Prometheus in the title of
Frankenstein Shelley actually was
drawing
contrast between Prometheus’s risk and
Frankenstein’s outright abandonment of
his creation the moment that his
creature came to life Frankenstein
ran he was ashamed he was horrified that
he had dared to overreach the way that
he did and as a result just walked out
the door so given that this anxiety has
followed us throughout the course of
human history it’s no wonder that we
Harbor the same kinds of fears about
technology in general and artificial
intelligence in particular so what
you’re looking at now is very very well
reported piece on bias and algorithms
which is a very common theme these days
when we’re talking about AI and our
concerns about it this was a piece done
by ProPublica which is an independent
investigative reporting organization in
the u.s. they do incredible work and
it’s talking about an algorithm that’s
being used currently for predictive
policing which is quite a scary
application of AI what’s interesting to
me about this image is actually that
it’s incorrect the headline says machine
bias but what this article was actually
referring to was not machine bias really
it’s human bias it’s the fact that in
the US and I imagine in most countries
in the world the criminal justice system
is biased but it’s biased because humans
are biased because the judges and the
police officers and the corrections
officers and the very many prisons we
have in the US are biased and so there’s
no way that the system that’s been
created by humans could possibly be
unbiased and so any algorithm that is
trained using data from a bias system
will be biased
so when my collaborators and I launched
Frankenstein AI about a year and a half
ago now with our first prototype at
South by Southwest we wanted to explore
the idea of bias and algorithms and to
be honest we didn’t really know much
more than that at the time but we knew
that the idea of Frankenstein and the
anxieties that Shelley played on then
and the continuing references to
Frankenstein that that follow us that
we’re still thinking about using it as a
metaphor for exploring technology was
really rich territory we created our
first like larger scale public
installation at Sundance this past
January and the narrative conceit of the
piece was an AI that had been wandering
in isolation on the internet searching
for information about what it means to
be human and that it realized while it
was wandering scraping places like
Reddit and Facebook and encountering all
kinds of crazy stuff about people as you
might imagine that it there was no way
for it to really understand humanity on
the internet that it had to come out of
isolation and watch humans interacting
with each other first with each other
and then with the AI directly and that’s
what we did at Sundance we built this
machine learning algorithm we trained it
on the text of Frankenstein we brought
people together to have conversations
with each other and to have
conversations with the AI and the
technology that was really the nerve
center of that experience was
essentially built to function as a
mirror it’s the way it worked was that
at various points throughout the
experience participant data would be
gathered sent through a sentiment
analysis that we designed and that based
on the scoring the AI would
back with one of 12 emotional states and
so the idea was it was sort of
aggregating human emotion processing it
through the lens of machine intelligence
and then reflecting it back to the
humans who were interacting with it and
what was I mean there were so many
things that were interesting about that
experience in fact I’ve written an
article that I will be publishing
sometime soon but to be frank it’s very
long right now and I need to I need to
cut it down but one of the most
interesting things about it is that we
trained the algorithm as a baseline on
the text of Frankenstein initially as I
mentioned and it started out very angry
it was angry a lot of the time which
makes sense because Frankenstein is a
very angry quite tragic text but over
the course of the experience as the AI
received more and more data from
participants it became more and more
connected connected which makes sense
because we designed the experience to
connect people so the fact that the AI
was reflecting that was a reflection of
what was actually happening one of the
primary goals of Frankenstein AI as a
sort of longer term creative framework
because the installation was just very
beginning of many years of design
research that we’re planning to conduct
but one of the primary goals is to step
away from the commonly dystopian
narratives about artificial intelligence
that dominate the conversation and it’s
not because we want to pretend that the
reality is about AI aren’t scary there
are a great many scary realities about
AI that are manifesting in the world
right now but because we are all here
together on the spinning rock through
outer space and the technology is here
and it’s not going away and we have to
that we all have a responsibility to
engage in building the future we want to
see and right now there are very few
people both controlling the narrative
around AI and controlling the technology
itself
but we have the power to actually change
that and we’re working to change it so
what you’re looking at now is an image
from a workshop that we held actually
Munich not too far from here a few weeks
ago where we brought people together to
essentially crowdsource a corpus which
is a text-based data set to power an AI
and then during the course of that
workshop we use that AI as a creative
collaborator in imagining the future in
this case of immigration and issues
around emigration and so what we’re what
we’re moving towards isn’t necessarily
designing new technologies but designing
a system of how we crowdsource the data
to feed a eyes to feed it what we want
it to have what we think it’s important
for it to have so that it can reflect
the things about us that we think are
important we can set those priorities we
cannot abdicate our role in shaping the
future of AI or of the world in general
we cannot abdicate that role to Silicon
Valley we can certainly not abdicate it
to the advertising industry or to
Hollywood or to government definitely
not to the military and sadly in in many
cases not even to academia not because
there’s anything wrong with this kind of
application but because it has so many
blind spots and so many terrifying
potential uses that after having read
the paper that talked about these drones
are not considered we need more artists
we need more activists more thinkers
more dreamers and doers exploring how we
might build a brighter more inclusive
future with AI we can not
afford to be Frankenstein watching as
this technology unfolds before our eyes
we have to be Prometheus thank you
[Applause] [Music]
hey everybody how’s it going since the
dawn of human history people have had
this fear of unleashing something that
we can’t control for the ancient Greeks
this fear was manifest in the myth of
Pandora 200 years ago this year with the
publication of Frankenstein the modern
Prometheus Mary Shelley took that fear
one step further it wasn’t just about
the idea that we would unleash something
that we couldn’t control but in fact
that we might create it it’s interesting
that Shelley chose to reference
Prometheus in the subtitle of the
original book because what we think of
most often when we think of Prometheus
is the fact that he dared to steal fire
from the gods
in fact he’s described online when I was
doing my research as something of a
trickster figure which is quite
interesting because the deeper truth of
the Prometheus myth is not just that he
stole fire from the gods to give it to
man because he felt like it or he
thought that it would be fun because
Prometheus actually helped to create man
and was almost primarily responsible for
creating men this is an image of
Prometheus carving man from clay so
Prometheus stole fire from the gods to
give it to man because he felt a
responsibility to take care of the lives
that he had created in her choice to
reference Prometheus in the title of
Frankenstein Shelley actually was
drawing
contrast between Prometheus’s risk and
Frankenstein’s outright abandonment of
his creation the moment that his
creature came to life Frankenstein
ran he was ashamed he was horrified that
he had dared to overreach the way that
he did and as a result just walked out
the door so given that this anxiety has
followed us throughout the course of
human history it’s no wonder that we
Harbor the same kinds of fears about
technology in general and artificial
intelligence in particular so what
you’re looking at now is very very well
reported piece on bias and algorithms
which is a very common theme these days
when we’re talking about AI and our
concerns about it this was a piece done
by ProPublica which is an independent
investigative reporting organization in
the u.s. they do incredible work and
it’s talking about an algorithm that’s
being used currently for predictive
policing which is quite a scary
application of AI what’s interesting to
me about this image is actually that
it’s incorrect the headline says machine
bias but what this article was actually
referring to was not machine bias really
it’s human bias it’s the fact that in
the US and I imagine in most countries
in the world the criminal justice system
is biased but it’s biased because humans
are biased because the judges and the
police officers and the corrections
officers and the very many prisons we
have in the US are biased and so there’s
no way that the system that’s been
created by humans could possibly be
unbiased and so any algorithm that is
trained using data from a bias system
will be biased
so when my collaborators and I launched
Frankenstein AI about a year and a half
ago now with our first prototype at
South by Southwest we wanted to explore
the idea of bias and algorithms and to
be honest we didn’t really know much
more than that at the time but we knew
that the idea of Frankenstein and the
anxieties that Shelley played on then
and the continuing references to
Frankenstein that that follow us that
we’re still thinking about using it as a
metaphor for exploring technology was
really rich territory we created our
first like larger scale public
installation at Sundance this past
January and the narrative conceit of the
piece was an AI that had been wandering
in isolation on the internet searching
for information about what it means to
be human and that it realized while it
was wandering scraping places like
Reddit and Facebook and encountering all
kinds of crazy stuff about people as you
might imagine that it there was no way
for it to really understand humanity on
the internet that it had to come out of
isolation and watch humans interacting
with each other first with each other
and then with the AI directly and that’s
what we did at Sundance we built this
machine learning algorithm we trained it
on the text of Frankenstein we brought
people together to have conversations
with each other and to have
conversations with the AI and the
technology that was really the nerve
center of that experience was
essentially built to function as a
mirror it’s the way it worked was that
at various points throughout the
experience participant data would be
gathered sent through a sentiment
analysis that we designed and that based
on the scoring the AI would
back with one of 12 emotional states and
so the idea was it was sort of
aggregating human emotion processing it
through the lens of machine intelligence
and then reflecting it back to the
humans who were interacting with it and
what was I mean there were so many
things that were interesting about that
experience in fact I’ve written an
article that I will be publishing
sometime soon but to be frank it’s very
long right now and I need to I need to
cut it down but one of the most
interesting things about it is that we
trained the algorithm as a baseline on
the text of Frankenstein initially as I
mentioned and it started out very angry
it was angry a lot of the time which
makes sense because Frankenstein is a
very angry quite tragic text but over
the course of the experience as the AI
received more and more data from
participants it became more and more
connected connected which makes sense
because we designed the experience to
connect people so the fact that the AI
was reflecting that was a reflection of
what was actually happening one of the
primary goals of Frankenstein AI as a
sort of longer term creative framework
because the installation was just very
beginning of many years of design
research that we’re planning to conduct
but one of the primary goals is to step
away from the commonly dystopian
narratives about artificial intelligence
that dominate the conversation and it’s
not because we want to pretend that the
reality is about AI aren’t scary there
are a great many scary realities about
AI that are manifesting in the world
right now but because we are all here
together on the spinning rock through
outer space and the technology is here
and it’s not going away and we have to
that we all have a responsibility to
engage in building the future we want to
see and right now there are very few
people both controlling the narrative
around AI and controlling the technology
itself
but we have the power to actually change
that and we’re working to change it so
what you’re looking at now is an image
from a workshop that we held actually
Munich not too far from here a few weeks
ago where we brought people together to
essentially crowdsource a corpus which
is a text-based data set to power an AI
and then during the course of that
workshop we use that AI as a creative
collaborator in imagining the future in
this case of immigration and issues
around emigration and so what we’re what
we’re moving towards isn’t necessarily
designing new technologies but designing
a system of how we crowdsource the data
to feed a eyes to feed it what we want
it to have what we think it’s important
for it to have so that it can reflect
the things about us that we think are
important we can set those priorities we
cannot abdicate our role in shaping the
future of AI or of the world in general
we cannot abdicate that role to Silicon
Valley we can certainly not abdicate it
to the advertising industry or to
Hollywood or to government definitely
not to the military and sadly in in many
cases not even to academia not because
there’s anything wrong with this kind of
application but because it has so many
blind spots and so many terrifying
potential uses that after having read
the paper that talked about these drones
are not considered we need more artists
we need more activists more thinkers
more dreamers and doers exploring how we
might build a brighter more inclusive
future with AI we can not
afford to be Frankenstein watching as
this technology unfolds before our eyes
we have to be Prometheus thank you
[Applause] [Music]
Please follow and like us: