Press "Enter" to skip to content

The Future Face of Artificial Intelligence | Matthias Fuhr | TEDxHochschuleLuzern


[Music]
my grandmother loved reading books even
in her 90s
Siobhan’s very clear in her head and
eager for literature unfortunately as
she got older she more and more had
reduced abilities to read the text with
her eyes listen to the radio with her
ears and even all the book in her head
even worse once she had something with
her back and she could not sit up
anymore she was trapped in her bed more
than one day of course she had some
alarming device but unfortunately she
completely forgot that to activate this
absolutely traumatic situation finally
led to her decision to leave her beloved
home for me it was very hard to see her
in this way because she always was self
determined person and more and more
dependent on others to follow her
interests I asked myself how can I
support her how can I make her life
better and what kind of shape will I be
in when I am in her age in this time my
interests in artificial intelligence and
machine learning started and the
question rised are there some tools
occasions or even intelligent machines
that will support me in the future to
stay longer at home at my grandmother’s
it and if yes it’s such a machine a
super intelligent Terminator like robot
that can answer any question and can
solve any task you asked for meaning it
is superior to any human being or is it
more this kind of very highly
specialized tool that something like
r2d2 you may know him from Star Wars
this highly specialized tool that is
designed for solving some very specific
very concrete questions and solve very
concrete tasks like r2d2 is designed for
repairing component components of a
flying spaceship or in other words is it
Marty’s general artificial intelligence
or Marty’s narrow artificial
intelligence as we often call it
actually I’m not alone with this
question often customers ask me Mateus
what is the future face of artificial
intelligence what is the impact to our
business well I’m not futurologist and
can’t make any prediction what’s in 10
15 or maybe 20 years but as a data
scientist with many years of experiences
in artificial intelligence and machine
learning and working with such kind of
technology on a daily basis I think
in the very near future we will be
surrounded by a nearly limitless number
of highly specialized tools and
applications so why do I come to this
conclusion I think I have mainly two
arguments on my side so first of all
applications that imitate human behavior
to an until recently unforeseen level
are already here for example take
inspection of an oil pump so just
download this on eliezer app from the
App Store hold it to an old oil pump and
paste on the sound spectrum of this oil
pump this little r2d2 will tell you
immediately what is the status of the
pump and it will make you suggestions
about how to optimize the energy profile
of this pump or just go to the
supermarket download app make a picture
with your mobile phone
and this little r2d2 app will
immediately tell you what kind of
product you have in front of you what
you can concrete it
what is the nutritional value and as
well you can find other people that also
like this kind of product in the same
category fall also all these intelligent
personal assistants like Google or
Syrian hey Siri what’s the weather
tomorrow in Lucerne or okay Google Place
and
male 2 my wife Francisca so this kind of
list here and examples can be extended
as long as one likes there are machines
that park your car machines that filter
spam there are machines that even play
some games and so forth and so on but
now you say how much is just
interconnect all these are 2d tools and
they will finally become a super
intelligent organism well actually I
think it’s not that easy because all
these are 2d tools don’t have access to
this big big world model in our head and
that is my second argument in favor of
the r2d2 world to show you what I mean I
adapted an example by a manual motion a
from Google so just pick a random
sentence or a question like this one
here out of your head
hey Siri a Google can I catch the train
for our r2d2 is why is it so hard to
properly answer a random question like
this one here so to answer a random
question like this one here properly our
r2d2 need to understand the deeper
meaning behind this question first of
all you need a concept about what a
train is so why do we catch train so is
train something we toss around and catch
like a ball no it’s something that
transports people
and why do we say caching why is it
ketching you know for me every morning
that is this mini deadline at the train
station in Zurich I run and 1015 seconds
too late and the Train is gone but in
Italy
so our cliche says that trains are
always laying and to be one minute late
at the train station it’s also okay so
you can also catch the train but on the
other hand when I lived in Japan once I
was one minute before departure in the
train but unfortunately this train
brought me to also come instead of Tokyo
sorry Mattias 20 seconds too early
correct platform wrong train believe me
I say you as a Swiss living in the
country of watches there is an
additional dimension of catching a train
in Japan so what are we talking about
what are we talking about we are talking
about this big big model of the world
that we all share together in our head
so this model consists of a big big
encyclopedia of facts we have some
historical dimension we even know we
have an idea about cultures we have a
sense for humor and so far so using this
world model together with our simulator
here we all here can easily resolve this
implicit references in this sentence
here from train to our world model
and from caching to our world model so
in order to properly answer any random
common-sense question on this planet our
r2t tools also need some kind of access
to this pick so then the great question
is how does this word model looks like I
personally think nobody really knows
then okay so what are the steps to go
towards this world model to finally
build it there are concepts there are
ideas but these concepts are still
subject to a research so do we have to
take the right technology do we have the
right methods to call this way to code
this way towards a world model to
finally build it I think for me
personally this question is still
unclear okay so we recognize that there
are many things to do before we finally
reach a general artificial intelligence
but what we can do in the meantime is we
can design a very highly specific
applications and train it with a lot of
data that can exactly answer this
question here but then you are back
again in the r2d2 world okay so let’s
summarize what is the face of artificial
intelligence in the very near future I
think it’s the r2d2 world we will be
surrounded by a nearly limitless number
of highly specialized intelligent
applications
so what is the impact of this world to
our society I think the impact will be
huge as already discussed on this stage
here about a year ago but just to
mention one thing the data generated by
all these are 2d tools can be used to
train algorithms in such a way that this
algorithms give the companies behind
these are 2d tools an immense power of
what kind of text we read online which
kind of book we hold in our hand and
which kind of music we listen in the
radio and which kind of rescue service
or doctor is called when we are as old
as my grandmother was entrapped at home
when our parents are as old as my
grandmother once they will be in a much
better shape there is a high probability
for this it’s up to everybody here that
they can enjoy their additional lifetime
in an self-determined matter not
dominated by some obscure interests
behind these r2d2 it’s up to us that we
defined the rules in such a way that
there is a fair competition and it’s up
to us that we will always have a choice
[Music]
[Applause]
[Music]
Please follow and like us: