so I’m told a degree in computer science
from
three seconds it’s possibly
so recently I picked up whose name I’m
fasting lottery counter revealed on
stage in TED talk on the back cover of
the book it made three bold visionary
headline claims about the future of
intelligent machines the first claim and
hold on to your seats for this because
it’s a big one
was that Oh in very very very soon we
could have robots that are as
intelligent as insects thank you
the second claim was a little bit more
impressive it said that in perhaps as
soon as five years we’ll have robots
that are as intelligence and as cats
which is pretty cool I think and that
means in perhaps three or four years
we’ll have robots as intelligent as dogs
the final claim sort of inevitably for
any kind of book about artificial
intelligence is of course the big one
which is that in perhaps as soon as ten
years we’ll have robots as intelligent
as people and like if you’ve seen the
news recently that seems increasingly
plausible regardless of your own kind of
personal opinions about this and what
you think sort of is you know whether
this is kind of a reasonable timeline
it’s kind of irrelevant because the
reality is this book was written in 1997
and I think it’s pretty fast to say that
the future that had been busy did not
sort of come to pass and in fact the
most intelligent robot in my life right
now is Rosie the robot which is so
intelligent that it navigates around my
living room by a system of colliding
with household objects and then changing
direction and then sort of diving under
the TV unit grabbing hold of cables and
trailing electrical equipment across a
low demand floor
so this isn’t exactly the future that I
was promised in the honest I was growing
up and but in spite of that in spite of
that disappointment were having to deal
with the sort of cavalcade of
futurologists and tech pioneers and
visionaries having us again and again
and again that the singularity is just
around the corner machines are going to
be more intelligent than human beings
they’re going to take out jobs they’re
gonna steal our wives and husbands and
yet increasingly less this kind of
growing disparity between the claims
that are being made and the reality that
we see in the world around us and to to
kind of explore that a bit I want to
take you on a little bit of a a thought
experiment now imagine and this is quite
exciting imagine you’re an office worker
in the year 1990 some of you have maybe
a little bit more easy than others but a
few things sort of become quite apparent
so it’s somebody graduating now this
would be a very different one and first
of all obviously growing black and white
this time it’s 20 and 27 nearly 30 years
ago if you have a computer on your desk
it’s not a particularly smart one it may
not be much more than a basic terminal
you’re probably looking at a user
interface that’s made out of ASCII art
if you want to send a message to a
colleague you’re probably going to have
to pick up the phone or send a memo if
you’re going to send a document to
another office you might be looking at
something called a fax machine but all
of the post or various other sort of
antiquated technologies so the world
that we have now pretty much doesn’t
exist even even as recently as this now
I want to take you through this kind of
experiment that was postulated by a
economist called Robert Gordon who did a
very similar sort of approach for
productivity during the 20th century but
imagine that you fall asleep like you go
out you have quite a heavy lunch down
the pub you come back to your office
you’ve had one or two Simoni you feel a
bit dozy and you fall asleep at your
desk and you sleep for ten years and you
wake up in the year 2000
somehow you’ve still got your job you’re
at the st. that’s even though
functionally useless for the last ten
years maybe you’ve been working on the
development unit for Windows I mean but
whatever the reason you’re still there
but the world that you’re in now is very
very different and this is kind of what
you’re faced with it’s basically the
same desktop I had when I started
University here in the year 2000
computers themselves have been
completely revolutionized we’ve had
Windows 3.1 Windows 95 windows 98
Internet Explorer we’ve got an email now
for sending documents mp3s solitaire
like this is a drastically different
Microsoft Office and boring PowerPoint
presentations like this one so this is a
drastically different kind of office
what the best of confusing while that
you have 10 years before and then
obviously this is going to be quite a
shock to the system so imagine that you
sort of go and have another drink and
you come back to your desk again and you
fall asleep you pass out and you sleep
for another 10 years now you wake up in
the year 2010 somehow you still want
your job in spite of being function of
the user for the last 10 years maybe
you’ve been working on the development
team for Windows 8 why you’ve been
asleep there’s been not one but two
major revelations in confusing the first
is the mobile revolution as kind of
epitomized by Apple we’ve had in that
time period the iPod the iPhone the iPad
so a whole range of the advices when you
got to the end of the nineties you have
Microsoft Office on your desktop
computer now you can for people using
your mobile phone it’s it’s an
incredible transformation and that’s
been matched by a parallel revolution
that’s happened in the world of the
internet so we have things like
Wikipedia reddit Facebook Twitter all of
those sites and services that pretty
much we use today are invented in this
same time period so again the world is
very very different in the space of 10
years then if we look at the time
since 2010
there’s this kind of weird sort of
slowing down in the race of changing the
stuff that we’re using this is a list of
some of the top websites and services
that we use online and you can see most
of them it’s actually pretty hard to
find anything from this decade that sort
of makes that top 50 because most of
this stuff has been around for quite
some time even things like Netflix or
whatsapp so that you think are quite new
or actually like 8 9 10 years old and so
it kind of begs the question if there’s
a major revolution happening right now
in artificial intelligence then where is
this manifesting in the world that we
see around us and you know this is the
decade where we’ve seen all the
superficial ease of quite major
achievements so 2011 saw IBM’s Watson
defeating their human champion at the
Jeopardy for the first time the year
after that was that are sort of dawn of
the the kind of deep learning era when
there were major breakthroughs in image
classification and natural language
processor and on the back of that in
2016 we had alphago beating a human
player at go while sorry the human
champion at go for the first time so
that you know there’s this weird sort of
dichotomy between these on one hand sort
of major advances and sort of challenges
being met and on the other hand this
sort of lack of this stuff filtering
through to the rest of the world and
even this year we found that with both
IBM and with deep mind there’s a kind of
growing skepticism about their ability
to take some of these technologies and
actually do some of the things that they
claim they’d be able to do in industry
in government in society and I could
spend all afternoon kind of it’s racing
through all the various legal ethical
technical data problems that these
companies have had but I would get
kicked off stage at some point but it
all kind of boils down to a single kind
of insight which is that it turns out
that it may well be the running a kind
of messy complex human system like the
NHS is actually a lot more difficult
than playing a board game now you may
say but that’s slightly unfair that was
sort of picking on the moon shots on the
really sort of you know high-end kind
abstracts and what about the sort of
steady improvements that are being made
by machine learning by AI and the tools
that we use every day but even here if
we look at some of the sort of major
companies in the field if we take
facebook for example this is an advert
that I was served last week for height
boosting shoes and to be honest that six
will one I think I’ve already got a
surplus of height if anything I need
much more and yeah you know Facebook’s
entire business is effectively
advertising in terms of revenue and so
you would think that if major advances
were being made there were filtering
through to their products this would be
one of the first places that we’d see it
and yet we still regularly served
adverts that gb/s our best most often
for things that you bought last week
Google has many of the same problems
when you look at the products and you
look at some of the decisions their
algorithms are not necessarily that
smart this was the top stories
highlighted during the Las Vegas
shooting last month and obviously 4chan
popping up in the results there with
some completely dodgy stuff but also
pressed Evo an Iranian propaganda outlet
so there’s this real sort of disparity
between the technology that the people
are talking about what we’re actually
seeing there are better examples and you
know people point to things like
self-driving cars for example or the
rise of voice-activated personal
assistants but when we start breaking
down the advances that make those
technologies possible it’s not
necessarily improvements in intelligence
that have enabled them often it’s things
like in the case of self-driving car
seat advances in lidar sensors and the
various other sensors that are used if
you look at Syria or Alexa actually
improvements in microphone technology
and miniaturization of helped to move
them faster than improvements in how
bases being processed um now all of this
sounds a little bit negative and it’s
not like I’m trying to suggest here that
machine learning is useless or the
reponses aren’t being made or the deep
learning doesn’t ultimately have the
potential to change the world but it
feels like we’re so busy banging on
about what might
than in 30 years or 50 years of 70 years
time that we’re not really giving people
a sort of sensible or fair assessment of
what’s happening right now and that’s
bad for investors it’s bad for customers
it’s bad for public understanding and
it’s bad for policy based on this stuff
and so in a way this is kind of like the
an seats head silk in the sense that a
lot of people like to come up and kind
of sell you an amazing vision of what
the world might be like in the future
and what I’m kind of saying here is
let’s stop banging on about the future
quite so much that maybe focus a little
bit more on the challenges and problems
that we have right now because there’s
plenty of them and well I’d say to
anyone if you’re worried that an AI is
going to steal your job in the next few
years then just take a look at some of
the products in the world around you and
I think you’ll find there still
reassuringly crap
[Music]
[Applause]
from
three seconds it’s possibly
so recently I picked up whose name I’m
fasting lottery counter revealed on
stage in TED talk on the back cover of
the book it made three bold visionary
headline claims about the future of
intelligent machines the first claim and
hold on to your seats for this because
it’s a big one
was that Oh in very very very soon we
could have robots that are as
intelligent as insects thank you
the second claim was a little bit more
impressive it said that in perhaps as
soon as five years we’ll have robots
that are as intelligence and as cats
which is pretty cool I think and that
means in perhaps three or four years
we’ll have robots as intelligent as dogs
the final claim sort of inevitably for
any kind of book about artificial
intelligence is of course the big one
which is that in perhaps as soon as ten
years we’ll have robots as intelligent
as people and like if you’ve seen the
news recently that seems increasingly
plausible regardless of your own kind of
personal opinions about this and what
you think sort of is you know whether
this is kind of a reasonable timeline
it’s kind of irrelevant because the
reality is this book was written in 1997
and I think it’s pretty fast to say that
the future that had been busy did not
sort of come to pass and in fact the
most intelligent robot in my life right
now is Rosie the robot which is so
intelligent that it navigates around my
living room by a system of colliding
with household objects and then changing
direction and then sort of diving under
the TV unit grabbing hold of cables and
trailing electrical equipment across a
low demand floor
so this isn’t exactly the future that I
was promised in the honest I was growing
up and but in spite of that in spite of
that disappointment were having to deal
with the sort of cavalcade of
futurologists and tech pioneers and
visionaries having us again and again
and again that the singularity is just
around the corner machines are going to
be more intelligent than human beings
they’re going to take out jobs they’re
gonna steal our wives and husbands and
yet increasingly less this kind of
growing disparity between the claims
that are being made and the reality that
we see in the world around us and to to
kind of explore that a bit I want to
take you on a little bit of a a thought
experiment now imagine and this is quite
exciting imagine you’re an office worker
in the year 1990 some of you have maybe
a little bit more easy than others but a
few things sort of become quite apparent
so it’s somebody graduating now this
would be a very different one and first
of all obviously growing black and white
this time it’s 20 and 27 nearly 30 years
ago if you have a computer on your desk
it’s not a particularly smart one it may
not be much more than a basic terminal
you’re probably looking at a user
interface that’s made out of ASCII art
if you want to send a message to a
colleague you’re probably going to have
to pick up the phone or send a memo if
you’re going to send a document to
another office you might be looking at
something called a fax machine but all
of the post or various other sort of
antiquated technologies so the world
that we have now pretty much doesn’t
exist even even as recently as this now
I want to take you through this kind of
experiment that was postulated by a
economist called Robert Gordon who did a
very similar sort of approach for
productivity during the 20th century but
imagine that you fall asleep like you go
out you have quite a heavy lunch down
the pub you come back to your office
you’ve had one or two Simoni you feel a
bit dozy and you fall asleep at your
desk and you sleep for ten years and you
wake up in the year 2000
somehow you’ve still got your job you’re
at the st. that’s even though
functionally useless for the last ten
years maybe you’ve been working on the
development unit for Windows I mean but
whatever the reason you’re still there
but the world that you’re in now is very
very different and this is kind of what
you’re faced with it’s basically the
same desktop I had when I started
University here in the year 2000
computers themselves have been
completely revolutionized we’ve had
Windows 3.1 Windows 95 windows 98
Internet Explorer we’ve got an email now
for sending documents mp3s solitaire
like this is a drastically different
Microsoft Office and boring PowerPoint
presentations like this one so this is a
drastically different kind of office
what the best of confusing while that
you have 10 years before and then
obviously this is going to be quite a
shock to the system so imagine that you
sort of go and have another drink and
you come back to your desk again and you
fall asleep you pass out and you sleep
for another 10 years now you wake up in
the year 2010 somehow you still want
your job in spite of being function of
the user for the last 10 years maybe
you’ve been working on the development
team for Windows 8 why you’ve been
asleep there’s been not one but two
major revelations in confusing the first
is the mobile revolution as kind of
epitomized by Apple we’ve had in that
time period the iPod the iPhone the iPad
so a whole range of the advices when you
got to the end of the nineties you have
Microsoft Office on your desktop
computer now you can for people using
your mobile phone it’s it’s an
incredible transformation and that’s
been matched by a parallel revolution
that’s happened in the world of the
internet so we have things like
Wikipedia reddit Facebook Twitter all of
those sites and services that pretty
much we use today are invented in this
same time period so again the world is
very very different in the space of 10
years then if we look at the time
since 2010
there’s this kind of weird sort of
slowing down in the race of changing the
stuff that we’re using this is a list of
some of the top websites and services
that we use online and you can see most
of them it’s actually pretty hard to
find anything from this decade that sort
of makes that top 50 because most of
this stuff has been around for quite
some time even things like Netflix or
whatsapp so that you think are quite new
or actually like 8 9 10 years old and so
it kind of begs the question if there’s
a major revolution happening right now
in artificial intelligence then where is
this manifesting in the world that we
see around us and you know this is the
decade where we’ve seen all the
superficial ease of quite major
achievements so 2011 saw IBM’s Watson
defeating their human champion at the
Jeopardy for the first time the year
after that was that are sort of dawn of
the the kind of deep learning era when
there were major breakthroughs in image
classification and natural language
processor and on the back of that in
2016 we had alphago beating a human
player at go while sorry the human
champion at go for the first time so
that you know there’s this weird sort of
dichotomy between these on one hand sort
of major advances and sort of challenges
being met and on the other hand this
sort of lack of this stuff filtering
through to the rest of the world and
even this year we found that with both
IBM and with deep mind there’s a kind of
growing skepticism about their ability
to take some of these technologies and
actually do some of the things that they
claim they’d be able to do in industry
in government in society and I could
spend all afternoon kind of it’s racing
through all the various legal ethical
technical data problems that these
companies have had but I would get
kicked off stage at some point but it
all kind of boils down to a single kind
of insight which is that it turns out
that it may well be the running a kind
of messy complex human system like the
NHS is actually a lot more difficult
than playing a board game now you may
say but that’s slightly unfair that was
sort of picking on the moon shots on the
really sort of you know high-end kind
abstracts and what about the sort of
steady improvements that are being made
by machine learning by AI and the tools
that we use every day but even here if
we look at some of the sort of major
companies in the field if we take
facebook for example this is an advert
that I was served last week for height
boosting shoes and to be honest that six
will one I think I’ve already got a
surplus of height if anything I need
much more and yeah you know Facebook’s
entire business is effectively
advertising in terms of revenue and so
you would think that if major advances
were being made there were filtering
through to their products this would be
one of the first places that we’d see it
and yet we still regularly served
adverts that gb/s our best most often
for things that you bought last week
Google has many of the same problems
when you look at the products and you
look at some of the decisions their
algorithms are not necessarily that
smart this was the top stories
highlighted during the Las Vegas
shooting last month and obviously 4chan
popping up in the results there with
some completely dodgy stuff but also
pressed Evo an Iranian propaganda outlet
so there’s this real sort of disparity
between the technology that the people
are talking about what we’re actually
seeing there are better examples and you
know people point to things like
self-driving cars for example or the
rise of voice-activated personal
assistants but when we start breaking
down the advances that make those
technologies possible it’s not
necessarily improvements in intelligence
that have enabled them often it’s things
like in the case of self-driving car
seat advances in lidar sensors and the
various other sensors that are used if
you look at Syria or Alexa actually
improvements in microphone technology
and miniaturization of helped to move
them faster than improvements in how
bases being processed um now all of this
sounds a little bit negative and it’s
not like I’m trying to suggest here that
machine learning is useless or the
reponses aren’t being made or the deep
learning doesn’t ultimately have the
potential to change the world but it
feels like we’re so busy banging on
about what might
than in 30 years or 50 years of 70 years
time that we’re not really giving people
a sort of sensible or fair assessment of
what’s happening right now and that’s
bad for investors it’s bad for customers
it’s bad for public understanding and
it’s bad for policy based on this stuff
and so in a way this is kind of like the
an seats head silk in the sense that a
lot of people like to come up and kind
of sell you an amazing vision of what
the world might be like in the future
and what I’m kind of saying here is
let’s stop banging on about the future
quite so much that maybe focus a little
bit more on the challenges and problems
that we have right now because there’s
plenty of them and well I’d say to
anyone if you’re worried that an AI is
going to steal your job in the next few
years then just take a look at some of
the products in the world around you and
I think you’ll find there still
reassuringly crap
[Music]
[Applause]
Please follow and like us: