Will robots take over the world? (24 Feb 2011)


Uploaded by UCLLHL on 28.02.2011

Transcript:
>> Thank you for that introduction and thank you
to Dan and the team at the lunchtime lecture
for inviting me here today to talk about one of my passions
which is robots and [laughter] somebody else's passion, too.
So, I don't normally talk without notes,
so I'm gonna do my best to try and describe what my research is
about and [corrupt audio] thing and some things not clear.
We can talk about it afterwards, or you can contact me
in the Department of Anthropology.
So, first of all, I chose this image very deliberately.
I'm gonna tell you more about it when we go through the talk,
but you can see this kind of metallic object,
RUR [phonetic] emblazoned across the chest,
and a worker kneeling down in front of it.
And someone has quite carefully placed the hand
on the head of this worker.
And so, you know, we kind
of for some reason have these psychic associations,
if you like, between robots and destruction and the end
of the world and things.
And so, what I'm hoping to do in this talk is talk about some
of the backdrops to that, why do we think about robots
in that way, but also begin to think about, well,
what's actually going on in robotics and do we need to worry
about them taking over the world.
So, this is a kind of object
that you're probably all familiar
with when you had the term robot,
but I'm gonna show you the very, very first robots.
These were the very first robots.
They were characters in a play
in the 1920s called Rossum's Universal Robots and they--
the play was written by Czech writer called Karel Capek.
And basically, these robots, you know, people tend to think
of robots as kind of cute cuddly toys or, you know,
Hollywood depictions kind of devoid of politics.
But the first robots were actually created and imagined
in a time of absolute political turmoil.
You just had the First World War, you know, it finished--
had a devastating impact across Europe and--
so people will kind-- and people are kind of reflecting
on what does it mean to be human, what makes us human,
those kind of question.
And this kind of context is what inspired Capek
to kind of write this play.
And interestingly, these robots being human, they are actually
in the play assembled on a production line,
a bit like the Ford manufacturing production line.
So, even though they are human, they are assembled
and these robots are designed to labor,
and that is their primary purpose in society.
So, then what happens, when Karel Capek writes this play
and it starts to be reproduced in Paris, London and New York,
other artists start to take this robot image and they start
to reproduce it in their own particular ways and they start
to interestingly turn this robot character into a machine.
And I love the fact that this robot is called Eric, you know.
We tend to think of Cybertron or, you know,
kind of funky names for robots,
but this one was just called Eric.
And it was made by Captain Richards in [inaudible]
and as you can see, it's kind of--
this robot was just a simple mechanical object.
It could stand up.
It could move its arm up and down.
So very simplified, you know, kind of movements.
But, you know, you can see, I mean, it looks incre--
I think it looks pretty incredible myself.
So, we have this kind of this robot becoming machinelike
which is starting to distance itself more and more and more
from this kind of human character
that was in the first play.
So, what kind of-- made the robot turn into a machine?
We got to remember around the turn--
the return in the century particularly, in the 1920s,
there was an obsession with machinery and production
and this idea that through machines we could be delivered
to these new societies, you know, so you had a competition
between the left in the Soviet Union and in parts
of Europe arguing that we could use machinery and technology
to kind of reduce people's labor time so that they could go
and be freed to look art and literature
and do whatever they please.
And then in the-- in America which was the heart
of capitalism at that--
well, there are lots of capitalist economists,
but it was certainly creating new models of production.
You have the Ford production line being developed.
And so in both left and right wing philosophies at that time,
there was this definite emphasis on production and this idea
that through the machine, we could like deliver
to this new society and in--
as well as the artists themselves became fascinated
with the machine.
So we have, you know, modern times the parody
of the modern production system by Charlie Chaplin.
In the middle we have this picture here which is
from Metropolis which is basically a big machine
with people working at machines and when they die they're thrown
into a machine that consumes them
and the process starts all over again.
So, this is kind of an idea as well that--
as well as all these new technology
that was liberating us all,
it was also dehumanizes turning us into things.
And, I mean, the picture on the right is just one example
of this, but every aspect of life was basically reimagined
in machine form, buildings were imagined as machine,
cities as machines, the body as a machine.
Everything was kind of--
the impact of machine modernism was extremely powerful.
And I suppose, in a way a machine modernism
in different forms still exist today.
So, this like-- this abs-- you know, the fact that this
and this robot becomes--
the robot becomes a machine absolutely horrified its creator
Karel Capek.
He says and he wrote this in third person for some reason.
It is with horror frankly that he rejects all responsibility
for the idea that metal contraptions could ever replace
human beings and that by means
of wires they could awaken something
like life, love, and rebellion.
He would deem this dark prospect to be either an overestimation
of machines or a grave offense against life.
You know, so a bit like the character in Frankenstein.
You know, Frankenstein creates his monster and he kind
of loses control of his monster.
Karel Capek was absolutely horrified that his kind
of robot was being turned into a machine and--
and the machine basically started
to dominate the imagery of a robot.
So, I don't know if anyone--
I'm sure you all are kind of had an idea
of a robot being machine when you came in.
So, let's have a look at the definitions of a robot.
These are just a few.
They're quite vague actually.
Robot is taken from the Slavic word rabota,
which means to work extra, drudgery, slave--
extra work and-- but it also is a word for work
in Slavic languages with no negative connotations.
But the specific meaning that Capek got
from the term was related to drudgery.
Obviously there are-- because science now kind
of reproduces robots, we have a kind of mechanical, scientific,
technological definition.
And then also we have a person definition about someone
who themselves seem to be lacking in original thought,
you know, who won't
like criticize among we called them a robot.
So, we can now-- we think about robots in relation to people
and things and I think we still carry that kind
of imagination with us today.
So, another exciting thing
about this talk is it's the 50th anniversary of the first use
of industrial robots and this was it.
It's a robot arm.
I mean, if you're going to some manufacturing companies,
not in the UK unfortunately.
But overseas, you'll see lots of robots particularly
in the car industry, and they tend to perform very
like precise and what-- routine functions.
If for example the robots will precisely in place
and the machine-- the object that they are engaging
with also has to be precisely in the same place.
And if either of them are out of sync for any reason then
that there are problems.
But these robots are very effective.
I mean, Japan is kind one of the highest in users
of robots per industrial worker.
And just the same the--
once again, the GM assembly line was a kind of inspiration
for this and the creators got the idea from science fiction.
So it just-- it shows
that fiction has always been an important backdrop for thinking
about the technology of robots and artificial intelligence
and those kind of themes, and this is a character
from 2001 Space Odyssey that everybody knows.
And interestingly, when the writers Stanley Kubrick
and Arthur C. Clarke were trying to think of this idea,
they actually talk to scientists
and asked them what the future would be like, you know, 19--
in the 1960s they went to them and they asked,
"What's the future gonna be like in 2001?"
And they said, "Well,
we're gonna have these super advanced artificial intelligent
machines like HAL."
>> And I was-- I was a bit concerned
because the other day I heard about a supercomputer Watson
and I thought, "Oh, my God, you know, maybe they're right.
We've arrived here."
And then I started to notice a striking comparison.
So, we're gonna talk about Watson and for those
who might not know, Watson was on a show called Jeopardy
and it's created by IBM and the creator say
that it can understand natural language and therefore it's kind
of more super than our current supercomputers.
>> Well, for 600.
Bang, bang.
His Silver Hammer came down upon her head.
Watson.
>> What is Maxwell Silver Hammer?
>> Yes.
>> Literary character APV, for 400.
>> His victims include Charity Burbage,
Mad-Eye Moody, and Severus Snape.
He'd be easier to catch if you just name him.
Brad.
>> Who is Voldemort?
>> Correct.
>> Alternate meaning, 600.
>> Okay. If you want to watch more
of that you can look on YouTube.
But Watson just eventually end up winning that competition
which is heralded as his great breakthrough
in artificial intelligence and I'm not diminishing it.
I mean, it's-- it is an incredible machine.
But some things about like how did Watson do it?
Well, first of all,
Watson received the questions in text form.
It cannot see or hear so it cannot--
it has to receive information and communicate it
in a very specific way.
Responses were based on probability in terms of--
it had large databases of information and then it had
to search through the database and come
up with probable responses.
The way that it kind of engaged with the podium was--
when it decided on an answer it could like press a buzzer,
obviously not the machine, but there was a sequence in place
which allowed it to do that
and then it converted the text into a voice answer.
So, I thought I would search just Google and look
for these Jeopardy answers myself and I did this week
and you can do this at home
if you've got nothing else better to do.
So, I literally-- Jeopardy works by you get the answer
and you have to guess the question.
So, I literally put in just in Google these three questions.
On the first two, I got exactly the same response as Watson did.
And the third one interestingly I did get height
as the third question, but because Watson--
because Google works by searching
for relevant information, now when you search
for that phrase you get the Watson machine.
So, if you see what I mean.
But if you imagine that within Watson's own database banks,
it's not contaminated in that sense
like people can skew the results in that kind of way.
So, should we fear robots and-- ?
Oh, I think I'm doing okay for time, so I'll slow down a bit.
I was a bit worried there that I was going a bit too fast.
So, we have obviously the terminator image
of a robot taking over the world
and wreaking havoc and destruction.
Now, this kind of idea as well about this kind of character
who destroys the world, you can find the origin of the idea back
in the earlier play in Rossum's Universal Robots.
So if anyone wants to have a read of it.
And basically, certain of future society where robots do all
of the human labor and the humans themselves become lazy
and disaffected with life and there is a particular formula
that is developed for robot, the robot then becomes more superior
and they become powerful because they've got, you know,
physical strength and they take over the world.
So, we can find the roots of the story of kind
of robot destruction back in the 1920s.
Now, interestingly when I went to robotic labs and I got there,
it was more like a robot crash than a robot lab.
And I didn't really understand the significance
of why robots are made in this way before.
But actually, the roboticists themselves are
so concerned sometimes about the way
in which their object will be accepted in society
that they started recrafting them in particular ways.
There's this whole kind of philosophy about how
to design robots so that they don't frighten people.
One of them is to-- if you make them like cute
and a bit adorable, then people aren't gonna be frightened
by them, that's one thing.
If you make them more childlike, then people will kind
of interact with them like they're small children
and that will mean you don't have
to create very intelligent robots because, in my view,
people start to do all the work when they interact with them.
The other interesting thing about robot design is
about whether something is machinelike or humanlike and--
I'm just gonna give you two robots to choose from
and I just want to see a raise of hands.
So we have Casper the robot here on the--
up here, so that's a very humanlike robot and we have--
let's compare it with, I don't know, Mertz robot here.
Alright, who prefers Casper?
[ Inaudible Remark ]
>> Right. Okay.
Who-- may be you don't prefer any robot.
Who prefers Mertz?
[ Inaudible Remark ]
>> You like adult.
Well, there you go.
So, already there is kind of judgment that we make
about the way that we craft certain objects
and the more humanlike they look, the more they tend
to freak us out basically.
And the less humanlike they look,
the more kind of we want them.
And I've seen this on time and time again
and I think there is some truth to this.
Is that in robotics, it's kind of called the "uncanny valley".
So, this idea that certain objects,
if they look too humanlike but they don't behave humanlike,
they tend to frighten us.
And so, the way that some robotocists overcome these
issues is by designing robots that look very machinelike.
Another interesting thing I found in the United States,
when I did research there, is that they're trying
to make robots genderless and raceless.
So, this is an-- I didn't find this in the UK actually,
but I did find it in America.
They are very keen on their robotic artifacts kind
of being neutral in this respect.
But interestingly, even though none
of these robots have a gender who would say they--
well, let me just-- we'll go for a neutral one
in the middle, Kismet here.
Who say-- who thinks Kismet looks like a girl?
A few. A boy?
More. Neither?
Ah, about half and half.
So, they're obviously doing something right.
So, I'm gonna show you a bit of robot video here
and I guess you weren't expecting this
when I was gonna show you lots of robots,
but you'd be very surprised--
[ Corrupt Audio ]
[ Foreign Language ]
[ Corrupt Audio ]
[ Foreign Language ]
>> Anyway, there you go.
Robot babies.
Whatever next.
But there is also another kind of-- as an anthropologist,
I'm very kind of interested in this idea.
While when you're going to robotic labs,
you'll find all these like childlike robots and I-- and I--
and the way that they're trying to imagine the relationship
between robots and person is more like parent and child,
care and child and, you know,
anthropologically I think this is very interesting
and it's almost like developing of,
extending of familial relations, if you like, to nonhuman things.
And I think that some of that plays over and over again
in the way in which these robots are designed,
in the way robotocists talk about them.
>> So, you know, we have the kind of idea
of the robot as destroyer.
We have these robotocists
who are actually making these research platforms trying
to kind of negate these ideas.
But why did Capek's robots lead the way
in creating narratives of destruction?
And as I said, you know, the robot and the play was risen
in a very kind of turbulent time
and so Capek thought there was too much obsession on production
at that time, that's why his robot is about--
is related to the concept of labor and that's why Marx
and Ford [inaudible] kind of relevant here.
Because Capek saw himself as outside those kind
of philosophies and he didn't want to have a society
where there was everyone who was obsessed with production
and this machine modernism of the time.
So, he wrote that-- he wanted to create this idea
that when the robots attacked people, we could be--
we could think, "Gosh, isn't it great to be human?
Isn't it wonderful to be human?"
So, he kind of had these humanist meanings
in that respect.
So, how's his goal come true to some extent?
Well, I think robots have always kind of had a place in fiction
and even when they go into domains like science
and technology labs and into industry,
the fictional aspect never escapes it.
It's always there.
It's ever present and therefore it's a kind of technology
that is intimately bound up with cultural concepts.
So, for that reason, we now have robots engaged, you know,
technologists building robots
but were only first imagined in fiction.
Even the term robotocist is from a short story by Isaac Asimov.
So, even terms that we use in technology are kind of drawn
from fiction and one thing that I've learned in terms
of my interactions with lots of different robots and people
who make robots is that robots--
the making of robots shows us how incredible we are
as human beings.
They really do something that we take for granted
like having a simple communication is extremely
complex reproducing a machine.
And so maybe Capek inadvertently is, you know, is showing us
that it's quite incredible to be a human being or an animal.
And robots also show how distinctive our
characteristics are.
And so, my recent research has been looking at an area
of robotics called social robots and my new research is looking
at therapeutic robots.
But interestingly, like even trying
to develop a social robot is marked by lots of problems,
so I'm gonna show you a social robot here.
>> Do you really think so?
Do you really think so?
Do you really think so?
Do you really think so?
Do you really think so?
Do you really think so?
Do you really think so?
>> Right. So, this robot is what was called the first sociable
robot and it is called Kismet.
It was made by researchers at MIT.
And as you can see, the kind of--
the idea here is that if you have a robot and you get it
to express certain facial expressions,
then it can communicate a lot of emotions.
There is no denying that when we look at that robot,
we think it's expressing lots of emotion.
However-- well, I think there are certain layers
of emotion that's a very kind of simpler way
to understanding emotion in terms of behavior,
in terms of when I'm surprised my eyebrows are curved,
you know, my nose goes up slightly and, you know,
my eyes widen but-- so a lot of these social robots
and designing sociable machines, they use these kind of ideas
when they're designing them.
They think, "Well, what happens when a person looked surprised?
What happens to their face?
Maybe we can get the machine to combine those expressions."
And then, you know, we've got a robot demonstrating surprise
and the same with happiness and the same with joy
and all those other things.
Well, they don't enjoy.
But the emotions like sadness, happiness, disgust,
surprise are kind of-- is the usual emotional range.
But I actually think social robots
or unhumanoid robots are a bit of a dead end in robotics.
And the reason why is, not that I'm particularly
against the idea of humanoid robots or, you know, I can't,
you know, engage with an artificially intelligent entity
but I think it's more-- they just don't work.
If you go to robotic labs, the robots break down,
they're too complex, the systems just don't work and people say,
"Well, isn't it this just
because where we are in the technology."
Well, I think-- I don't just think it's a question
about the technology.
I think reproducing another person or another human being,
should I say, the body and the person
in the machine form is just an impossible task.
So I don't think the future is in humanoid robots
and I certainly don't think it's in social robots,
but I do think it's in other areas
and these are certain areas that I would like to see developing,
obviously not military robots but, you know,
these areas have applications.
Then we have nonhumanoid domestic robots like the Roomba.
I think that's a good application for robots.
The Roomba cost about 400 pounds
which is why I haven't bought one and it doesn't do stairs.
So you have to buy a robot Anna Hoover if you wanted
to do the whole house, unfortunately.
But maybe they'll get around that.
Educational robots.
I do a lot of outreach work and children absolutely love robots.
So, if anyone does outreach work and you're a scientist,
children really do get inspired by robots.
So, that's certainly an area I'd like to see developed.
Therapeutic robots.
Well, my current research project is looking at the use
of robots as potential aids to help children
with autism spectrum conditions.
And I just like-- give you an idea
of what the research is about.
It's not about encouraging a child to interact
with the robot, but just by having a robot in a room seems
to encourage an interaction between child
and other participant.
So, the robot seems to act as a mediator in this respect,
and I think that's a very interesting area of research
where robotics could develop.
But the research on this is all very new at the minute
so nobody really knows that the robot is effective or not.
Medical robots obviously, you know, another area already
in different kind of research institution.
You're seeing the use of medical robots and space robots.
I mean, I go-- I was kind
of having a conversation the other day with a friend
and we thought that on the surface
of Mars it's probably full of dead robots who have, you know,
just ran out of steam and-- but there's still a good way to kind
of explore the worlds, other entities out there.
So, these are areas I think in robotics
and I'm certainly taking-- I'd like to take my research
in the direction of these different types of areas.
So, I'm gonna end it there
but I'm gonna show you one last video and, you know,
the social robot has--
people are imagining that in the future we're gonna develop these
partnerships with machines and they could be like companions
for example for communities like the elderly community.
You have-- often like say one
of their biggest problems is loneliness.
So, there are some people out there who kind of want to kind
of develop technologies to be companion--
companion species for different groups in society.
Now, we don't have enough time to go into the ins and outs
of all that, but apart from the video that we often see
about robots and I'm sure there are some scientists here
but I'm sure they will agree
that when you see video robots working it's often robots
that have been videotaped a number of times
and they've finally got done what you wanted them to do.
So, whenever you see, of course, the surprise then enjoy it,
but when you see robot video,
I would say always be a bit cautious 'cause it probably took
hundreds of hours to get to that stage.
So, this is a fantastic robot.
It's one of my favorite robots called Mertz
and the reason why I like this one is 'cause the researcher,
right, decided not to do this, decided not to just put it,
you know, videotape it, put the video on the web and like
to say how great robots were.
>> She actually put her robot in a public space in--
at MIT and she let people interact with it.
And I think-- I think this interaction between robot
and the-- and the people just shows--
well, it's quite funny but it-- it's-- it just shows that--
actually it's not so easy just
to create a few facial expressions to engage people
in a social interaction.
So, I'm gonna end on there and just show you--
>> Hello. What is your name?
>> Oh, no.
Sorry. This one.
I want to make it loud there but-- anyway.
>> Hello. What is your name?
[ Laughter ]
>> My name is [inaudible], my friend.
What is yours?
>> Your name is [inaudible].
>> Is what?
>> Yes or no?
[ Laughter ]
>> No, [inaudible].
>> Say yes or no [inaudible].
[ Laughter ]
>> No.
>> Please look at my face.
Uh.
>> He was speaking in French.
Are you speaking French?
>> Say yes or no.
>> No.
>> Hello. What is your name?
[ Laughter ]
[ Inaudible Remark ]
>> Martha.
>> Your name is [inaudible].
[ Laughter ]
>> Yes or no.
>> No. It's Martha.
>> I cannot see you.
>> I'm here.
Hello.
>> Please look at my face.
>> What? What did you say?
>> Green face [inaudible].
>> What?
[ Inaudible Remark ]
[ Laughter ]
>> Green face [inaudible].
>> So ridiculous, you know.
>> Answer correctly.
>> Hello. What is your name?
[ Laughter ]
>> Martha.
And yours?
>> Please do not move too much.
>> You move too much.
>> I'm here.
>> I cannot see you.
[ Laughter ]
>> Can you see me now?
>> Please look at my face.
>> Are you seeing me now?
>> Green face [inaudible].
>> Huh?
>> You are to follow me.
[ Laughter ]
>> No. I'm not following.
What do you want?
[ Inaudible Remark ]
[ Laughter ]
>> No, don't get mad.
[ Inaudible Remark ]
>> Let's go and talk.
>> Too many words.
>> Oh, too many words.
>> Okay. I don't understand any language but I am trying
to learn and repeat simple word you say.
>> Come closer please.
>> So, you cannot communicate.
>> Come closer please.
>> Am I close enough?
>> You are too far away.
[ Laughter ]
>> Come closer please.
>> I think he's [inaudible].
[ Inaudible Remark ]
[ Laughter ]
>> Okay. Well, just on that.
It could have gone on forever
but you can see how much people worked hard
to have an interaction with that robot and I think it's great
that the researcher like put her work out there,
put it in a public space and let people just interact with it
and I think this probably gives us the best--
this is the most natural and best account of what it's
like to interact with the social robot.
Okay, thank you.
[ Applause ]
>> Thank you Dr. Richardson.
It's really a fantastic subject to explore today.
We have about 3 or 4 minutes for questions.
If anyone would like to post a question, we got two microphones
on either side of the room and we--
if you could just hold off 'til you start speaking 'til the
microphone gets to you.
Do you have a question?
We got one here.
>> This question isn't very well formulated but I think it's
of interest that the idea of a humanoid robot was like sort
of invented first a long time before there was any possibility
[inaudible] having a humanoid robot.
And you mentioned the feeling of disgust that we have
at an attempt to make a robot humanoid.
I mean that last video was very humorous and comical,
but earlier on you mentioned that there was this sort
of feeling of disgust if you try to make a robot humanoid.
And I think there is probably something deep going on there.
I'm trying to think we-- that a law should be passed saying
that you shouldn't be allowed to make robots humanoid
and maybe we should just talk about computerized machines
or interactive machines or something of that nature.
>> Thanks for your question.
I don't think there should be laws passed.
I mean technology should be [inaudible] whatever they like.
And what I was kind of saying was that when you build a robot,
the more humanlike you make it,
the more it kind f provokes anxiety in us.
So actually, if you create the humanoid
that look very machinelike,
people will accept it a little bit more and--
but then that raises questions
about whether you can build a humanoid,
whether you should build a humanoid,
whether resources should go in that area in robotics.
And interestingly, the west is always contrasted
with like Japanese societies and we're told
that Japanese people love robots and--
I have a Japanese informant
who tells me not every household in Japan has a robot.
And recently, that started--
companies have started moving away from humanoids
and toward thinking about-- precisely those things,
you know, other type of intelligent agents
that could help disabled communities, the elderly
or just, you know, enrich our lives in different ways
and I think there's definitely a move away from the humanoid.
It was kind of the 10 years, 2000, 2010,
I mean there are still some humanoids around
but I think we're coming to end of that now.
>> [Inaudible] we could touch a bit more
on one aspect that's robots of the future we'll have.
First of all, return that very primitive machines there
and your first leader was military robots.
There will be a lot of spin-offs of those.
One of the things they will do is communicate with each other
in networks in a way that we as humans can't.
We are limited by our language, we are limited
by the speeds of our brains.
We are talking potentially about some quite ominous technology
and whether the front end is humanoid
or whatever form it takes, I think the important thing
to remember is that a network of robots of whatever kind,
intelligent robots would have an amazing form of communication
that we can't even envisage.
And I think there's a caveat there to talk
about robots taking over the world and the way
that science fiction portrays it is probably not what's gonna
happen but what could well happen is that they will--
computers already made this redundant.
The pocket calculator made people forget how
to do simple calculations in their heads, that was the start.
I can see a point where if we're painting a science fiction,
the 1984 caveat, shall we say,
of robots is don't let them takeover and we should remember
that because there is-- there is
that real potential if we're not careful.
>> So, you think they might take over.
[ Laughter ]
>> Yes, absolutely.
Absolutely.
>> See, I mean, I think what artificial intelligence actually
tells us is that machines are very good at certain--
doing certain kinds of things.
They are very good at doing logical things,
mathematical things, things that require searching--
searching primes, sorting through lots of information,
I mean, you know, those kind of things.
But when it comes to--
[ Inaudible Remark ]
>> Well, actually that's--
the mathematics of defining this room
so you could have a robot navigate a space is extremely
complex, and I don't think you could.
And even if you could put all the parameters into this room,
all that someone has to do is put a chair there and that's it.
Your kind of calculations, you know, don't--
aren't effective anymore.
So, I think-- I'm not saying there aren't great areas
of research.
I mean, it's wonderful that some
of these interesting developments and obviously
in technology but, you know, I do think it shows how distinct--
distinctive we are as human beings.