think about human like machines well
probably this to what do we imagine when
we see this word well well my favorite
all-star something very interesting to
say about a definition of robots so
Douglas Adams in his book The
Hitchhiker’s Guide to the galaxy had an
interesting definition of a robot he
wrote that the Encyclopedia Galactica to
find a robot as a mechanical apparatus
designed to do the work of a man he also
offered us another definition he wrote
that the marketing division of the
Sirius cybernetics corporation to find a
robot as your plastic pal who’s fun to
be with so these two alternative
definitions really show us that there
are a fundamental dichotomy in our
perception of robots on one hand we’ve
got the robot of the future the
purported robots of fiction here they
look and they act very much like us the
really stereotyped has a basically a
slightly naive metallic servant on the
other hand we’ve got the robot of
reality now their form is largely
dictated by their function and you
really don’t resemble us very much at
all now earning personal perception of
robots is probably more limited so
perhaps we have a robot that can vacuum
or flaws or maybe we will ask a lexer to
change our music to switch a light on
maybe even orders pizza soon we have
robots that will drive our cars and
maybe one day the dream of personal
robotics in the house although this is
just been round the corner over the last
40 years now so who knows what the
future of robotics holds I don’t know
I’m not a roboticist I’m not here to
talk about a new dawn of ubiquitous
robotics I’m a psychologist I’m
interested in what makes us trust these
new machines now I’m sure most of us
would trust a robot to sweep our floors
or to change the music but how many of
us would trust a robot to drive our cars
and how many of us would trust a robot
to look after our children or elderly
relatives so that’s why I’m particularly
interested in robot interaction because
it really drives how which how we trust
them so
what I’m interested in looking at is how
we do interact with robots playing games
now games are a particularly interesting
safe framework for looking at robot
interaction because it’s something we’re
all familiar with we all have experience
with games even if it is limited to
angry games of monopoly at Christmas
with our families but games offer us an
interesting way to look at how we
interact with robots as partners and
also as opponents playing against other
humans and robots as partners or
opponents for example in this picture
here we can see our female participant
is partnered with the blue robot and
what they have to do is come up with a
common strategy to use in a game against
their opponent the red robot robot in
the corner now the advantage of using
games is that our decisions and our
interactions basically provide us with
an implicit and continuous measure of
trust as the game unfolds so it’s
through small tweaks modifications in
the rules of the game and also in the
behavior robots that allow us to explore
what allows us to trust these robots
when their opponents or their weather
opponents or they’re our friends so what
I try and do is look at this experiences
with robots and try and understand how
humanity the perceived humanity the
robots affects things because in games
for used to playing games against say
artificial virtual opponents and also
humans so you can easily look at that
dimension now the robots you saw in
there in that previous example were now
robots they’re widely used in
experimental robotics and they’re cute
small human-like robots that really fit
within our stereotype of what robots
should look like in the future now these
anthropomorphic robots on some big of
our searches for human-robot interaction
so instead of home to learn some new and
complex way to interact with these
machines then we’re invited to treat and
interact with them as you would our
fellow humans using language and social
skills we require and use in everyday
life so this was supposed to make the
robots easier to use a more intuitive
but they also offer another
advantage their humanity in their
familiarity are supposed to breed and
engender more trust in them more trust
say in this robot here than this
alternative which disc is oz g5 robot so
what we wanted to know is whether this
anthropomorphic trust dividend we could
see this effect within our games so what
we did is think about what sort of game
could be played which really reveal how
much we trusted those robots so let’s
think about this for a moment what game
would you play well say one of these
robots asked to borrow ten pounds from
you would you would you lend on the
money so you know give a gifting money
or lending money to a stranger on a
promise that could be no greater sense
of trust than that so effectively we
played these sorts of games with the
robots in the investment game but our
participants were given us some of
virtual money say 10 pounds and you’re
invited to vest some or all of that
money in one of these robots now when
they make the investment hopefully they
expect the robot will return the money
and potentially an extra sum of money is
a nice profit so how much money you
invest in these robots really shows how
much we trust them so in a resultant
experiment we found that people invested
more money in the human-like robot than
the less human like skittles robot now
in this game both robots behaved as we’d
expect from our you know from the
fictional companion like robots that we
ordered all know about they were
trustworthy they all consistently gave
us back more money than we gave them
however what if we change that what if
we experience robots aren’t quite
cooperating with what we want to do what
if they’re not going to do exactly what
we tell him to do so this is what we
tried to explore in this experiment on
one hand we had nice robots they again
gave you back more money than you
invested on the other hand we had nasty
robots they only occasionally gave you
back more money they invested and most
of the time they will give you back less
so they were ripping you off so in a
previous experiment we looked at
compared robots that looked more or less
like us this dimension of ads
for morphism but this is only one
dimension of this effect also we want to
look at how they behaved so in this
experiment we just use one type of robot
these human-like now robots but we
changed how they behaved on one hand we
had the human light now robots now they
moved and they gestured while they were
playing the game they spoke with a nice
warm human-like voice and they held our
gaze they tracked us as we looked around
in the game on the other hand we had
this synthetic type now and they stared
fixedly ahead unmoving and they spoke
with a very synthetic grating voice so
how did these changes to the behavior of
these robots affect how we trusted them
well what we found is that people
invested more in the night robot than
the last one there’s no surprises here
if you consistently invested in the
nasty robot then you can end up with a
lot less money than you initially
invested in what people didn’t shun the
nasty now completely they still risk a
little bit of their money each round in
hope they would see some more returns
what’s interesting is that they invested
consistently more money when the now
nasty now robot actually like a human
than it when it acted like a synthetic
machine on the other hand when now is
for being nice people didn’t really seem
to care about how the robot acted
there’s no real difference between the
synthetic and human right now in fact on
some measures of trust they seem to
prefer the synthetic robot to the human
like one so what can we take from these
two experiments well it seems that when
the robots are cooperating with us when
they’re doing our bidding we don’t
really care whether they resemble us too
much or whether they behave like us in
fact all might think that we might
prefer the robots to act more like
machines and to display the attributes
we we usually associate new machines
they’re predictable their continuous
they’re always going to do the same
thing and when they’re doing something
we want that’s what we want to continue
when the robots are less cooperative
when they start to turn against us
that’s when we looking for a human face
because the humanities received humanity
the robot gives us hope it gives us hope
that they will show some empathy and
through that empathy that might change
their behavior and be more like us so
the take-home message is this when we
start to see robots with human faces
around us be aware they might not always have your best intentions at heart