artificial intelligence is it something
like this or maybe this or something
more like this artificial intelligence
or AI is a subfield of computer science
which enables intelligence exhibited by
machines and even experts disagree about
what exactly defines AI but usually they
can agree that if something exhibits
intelligence if it interacts with its
environment in complex ways then we can
generally call it AI how many of you
thought of this when I said artificial
intelligence I work in the White House
in the Office of Science and Technology
Policy and the hand on the left belongs
to our President Barack Obama the hand
on the right belongs to a paralyzed man
named Nathan Copeland who’s using a
brain computer interface to actually
control and feel that robotic arm with
his mind the Office of Science and
Technology Policy works on some pretty
amazing things on behalf of the
president the US government and the
American people I have people in my
office who are working on policy matters
related to energy and the environment
cyber security nanotechnology
biotechnology and much much more
and my policy portfolio focuses on
artificial intelligence and robotics so
I get to wake up every morning and craft
policy related to AI machine learning
and technology platforms like automated
vehicles and drones you can think of
this talk steward of the schoolhouse
in 2016 the White House Office of
Science and Technology Policy started a
policy process around artificial
intelligence called the White House
future of AI initiative we hosted a
series of public events five across the
country and we solicited public input on
all different topics related to AI back
in October of 2016 we published a
significant public report on the future
of artificial intelligence and a
companion national AI research and
development strategic plan the almost 70
page AI report was a big deal primarily
because it’s really the most significant
thing the White House has ever said
about AI and it’s not the first time
that the White House has awoken to the
issue of artificial intelligence
obviously this technology has existed
for a while now though though the decade
sorry though though the field of AI has
existed for several decades recent
progress in AI has been driven by three
primary factors that have built upon
each other the availability of big data
which in turn has helped power more
sophisticated algorithms and machine
learning techniques and in turn has
relied upon faster computing
infrastructure policymaking is a process
of radical collaboration and the AI
policy process for us was no different
it is a synthesis of of complex ideas
both drawn from within and outside of
the government community and across the
country we had thousands of people tune
into our AI workshop series we spoke to
people across the public and the private
sectors experts in a variety of
different fields and people from all
the public the White House heard about
artificial intelligence and it was via
meetings phone calls emails and even
faxes yes really faxes from Americans
constituents all across the country and
even across the globe the Obama
administration heard about artificial
intelligence and today I’m gonna talk
about a few key takeaways from the
lessons that we learned the first is
that we have an imperative to use AI one
of the most common things we heard from
people whether they were public citizens
or experts in the field was that it is
critically necessary that we responsibly
develop and deploy as much artificial
intelligence as we possibly can
Jason Furman the chair of the
President’s Council of Economic Advisers
said that his biggest worry about AI was
that we don’t have enough of it other
experts argued that people are actually
dying today because we aren’t deploying
AI in settings where we know it can help
save human lives it’s not AI that will
bring on at the end of humanity they
said but the absence of these
technologies that is already killing
people and costing lives citing you know
medical errors and ineffective disease
treatment as areas where we can apply
machine learning techniques to save
people’s lives in health care ai is
helping us predict treat and and help
develop patient care in one example AI
is being used to process vast amounts of
oncological data to create a complete
map of pancreatic cancer so that we can
better target these processes with
treatment mechanisms that can more
effectively enabled cancer treatment ai
is also helping us save our planet
researchers are creating an autonomous
computational system called wild book
which helping biologists track and
evaluate animals for conservation
purposes across the globe using image
recognition techniques and crowd-sourced
publicly available images of all sorts
of Wildlife across the world we can
actually identify and track individual
animals like zebras elephants and whales
and this can help us reduce the
incidence of illegal poaching it can
help us conserve endangered populations
and can also help us better learn about
the species with which with which we
share our world and importantly a I can
also help us impact vulnerable
populations by providing critical
assistance in social challenge areas
like poverty and education in one recent
example a school district in Tacoma
Washington partnered with a tech company
to develop a machine learning model that
helped analyze student and historical
data to predict which students were most
at risk of dropping out of school and
prompt early intervention and after a
multi-year pilot of that system they
actually were able to boost graduation
rates from 55% to 78 percent in just
AI is impacting all areas of society and
it’s clear from these examples and many
others that that’s the case we also
learned that it’s critical for everyone
to develop and benefit from AI it’s a
pretty normal day if I am one of the
only woman in a lot of my meetings and
that story isn’t new not in the tech
sector and certainly not in the field of
computer science unfortunately the field
of AI is even more homogeneous today
just 18 percent of graduates and
computer science are female down from a
peak of 37 percent in 1984 and the field
of AI is even less diverse at a recent
neural information processing systems
conference in 2015 which is one of the
largest conferences on AI research in
the world just 13 percent of
participants were female blacks Latinos
and members of other ethnic and racial
minority groups are also severely
underrepresented compared to their
shares of the u.s. population in stem
and the tech workforce and certainly in
the field of computer science and the
subfield of AI as machine learning
techniques become more ubiquitous and we
use them more and more in everyday life
to replace human decision-making
it becomes increasingly important for us
to get this technology right last year
users of a popular photo app which
automatically applied tags to digital
photos and albums discovered that it was
actually classifying images of black
people as gorillas machine learning
often reflects biases that exist in
everyday life in this case it was
fundamentally a data problem but it also
impacts algorithms and the outputs of AI
systems it is critically important and
we all have a responsibility for us to
ensure that AI systems reflect diverse
populations and that we help reduce the
negative impacts of narrowly focused AI
development and unintended biases in AI
systems the third lesson that we learned
is that understanding AI can be hard but
is deeply important I don’t know if any
of you have seen this video of a robot
named Atlas by Massachusetts firm Boston
Dynamics but for those of you who
haven’t it depicts a robot stumbling
through a snowy field with dexterity
than somebody like me probably could and
getting pushed over by a person only to
spring back up again with an almost
disturbing resilience when this video
came online last winter I got a call
from my grandfather who had never before
talked to me about policy matters
related to my job what are you doing
about this robot he said so it had
personally affronted him somehow from
the computer screen on which he was
watching it I think this video was an
important lesson for a lot of people
because it made especially salient the
current state of Technology but it was
also a critical lesson for me in my
position at the time because it also
helped me realize the potential
disconnect that exists between the
people working on the science and the
policies behind these technologies and
everyone else and it turns out that
people vary in how well they understand
AI in a recent global consumer survey by
Weber shanwick and KRC research over
under 20% of
global survey participants actually so
that they understand a lot about AI and
over 30 percent actually admitted to not
knowing anything so you’re not alone if
you don’t really understand this field
of Technology interestingly though
despite of this we are pretty positive
about the overall impact of artificial
intelligence forty five percent of us
think it’s gonna be positive and only
seven percent of us think it’s gonna be
negative and fifty two percent of
consumers believe that AI will have a
positive impact on their lives
personally the fourth lesson that we
learned is that agile government is
critically important to the responsible
over the course of the development of
the AI report we learned a lot of things
but this was perhaps the most salient
there are many areas where it doesn’t
seem like government has direct
influence over AI as it’s related to
governance issues but this is actually
not the case today the government’s
making policy decisions about AI that
actually impact the way that this
technology is developed and will be used
in every everyday circumstances like
people like me and you and there’s no
area where this is more relevant than in
intelligent transportation which is an
area that my policy portfolio focuses
intently upon last month we had US
Transportation Secretary Anthony Foxx
come in and talk to us about intelligent
transportation platforms and when asked
how many questions he had received over
the course of his confirmation hearing
on either automated vehicles or drones
the answer he gave was zero and that was
three years ago
in 2013 this highlights the critical
importance that policymakers keep in
mind the current state of technology and
keep up with the pace of it in
developing policy and regulatory issues
related to it
neither issue was on the map he said my
concern has been that we’re moving into
the Jetsons era but we have a flint we
have Flintstone approaches to authority
and regulation and we can’t go the
distance with this until we really think
about things differently transportation
policy is something we took a deep look
at when we were writing the
administration’s AI report and there are
several reasons for it the first is that
transportation affects everybody
if you ever drive in a car or have
written an aircraft
get from point A to point B if you are
an owner or operator of one of the seven
million drones projected to be in US
airspace by 2020 this policy area
affects you when we were developing the
AI report one of the biggest lessons as
I mentioned was that we keep in mind the
pace of Technology automated vehicles
have the potential to save almost all
the 35,000 deaths caused a year by
traffic accidents and to totally
transform the way that we organize our
communities and allocate our time
autonomous aircraft systems are already
changing the ways in which we observe
and conserve our environment helping us
save lives and inspect critical
infrastructure it’s increasingly
important as technology helps enable
these critical social changes that we
help enable it equally from the
perspective of the government I want to
end with a focus on one critical
question which informs everything that
we do it’s scrawled on a whiteboard in
our office actually and that question is
how does they’re going to keep the
future in the present tense at the
beginning of this talk I spoke of the
notion of radical collaboration which
existed between policymakers and the
public community that we are affecting
and I want to underscore how critical it
is that policymaking remains a
team-building process that means that
everyone has to have a seat at the table
it’s more than just bureaucratic
optimism that I’m saying this we
literally cannot do it without you
we need academia industry and experts in
a variety of different communities to
step up and provide leadership and
educating the American public and the
next generation of innovators about the
importance of AI for good
we need the advocacy community the media
and people from a wider range of other
backgrounds to hold developers
accountable to AI systems so that they
responsibly represent the needs of a
diverse variety of populations we need
more woman and underrepresented to STEM
fields and the AI disciplines and we
need more people to reflect more often
about how these technologies impact
policymaking on a regular basis we need
all of you artificial intelligence
has incredible potential if only we all
work together to unlock it now more than
ever before
we need people from all backgrounds and
all disciplines to show up engage with
each other in with their government and
help us craft a more inclusive
forward-leaning future which helps move
us all in the right direction
thank you [Applause]