
computers this may sound funny coming
from somebody was a computer science
professor for almost thirty years but
it’s true at some point in my life my
parents decided that I should get a
university education so I had them sign
me up for computer science which I
figured would be useful and honestly the
first year I didn’t pay too much
attention but then it grew on me and
when at some point I discovered
artificial intelligence or AI I got
hooked I still am hooked and it’s not so
much because of self-driving cars or
personal digital assistants even though
they’re terribly important I want to
start by asking you six questions about
what you think computers could do
specifically I want you to tell me
whether in your opinion computers could
think understand be creative have
feelings have free will and be conscious
I’m not talking about today’s laptops or
even the giant computer clusters of
Google I’m speaking about computers in
principle how they might be in 50 or 500
years imagine that you had to vote on
these questions you had to answer yes or
no you couldn’t abstain you couldn’t
hedge and say well depends what you mean
how would you vote let’s do an
experiment
I’m going to ask you about whether
computers can think and you have to vote
thank you who’s a No
raise your hand thank you about 75% of
you think yes and a lot of you didn’t
raise your hand and I’ve got your names
let’s do another experiment um who
thinks and you have to vote that
computers could in principle have
freewill if you were a yes raise your
hand thank you
all the knows raise your hand all right
now I would say about 5% of you think
computers could have free
we’ll these are six questions that they
used to ask at my class at Stanford for
many years and I’d asked it twice at the
beginning of the course and at the end
and what invariably happened was that
after we had discussed various aspects
of artificial intelligence or AI and
what it meant to compute the phrase of
course computers can’t tended to
disappear people were less sure of
themselves and the vote at the end of
the course was much more favorable to
computers now in 15 minutes I don’t
think we’ll achieve quite a
transformation but I do hope we walk
away with a new perspective and it’s not
only a perspective about computers it’s
also impressed mainly a perspective
about ourselves as you know AI is the
science and technology of getting
computers to do smart things that we
usually associate with people things
like playing chess or speaking in
English or diagnosing a disease telling
a joke what do we know about AI in early
2017 first of all we know that everybody
speaking about it from the New York
Times from the White House to Dilbert we
also know that AI is hugely successful
especially in an era called machine
learning computers can recognize cats
and faces they can drive cars they can
play recreational games extremely well
beating world champions
we know that industry is betting heavily
on AI we also know that people are
worried and there two kinds of worry the
first kind is about negative
implications of AI on society mostly
unintended implications if cars can
drive themselves does that mean that
thousands of drivers will lose their
jobs will biases that exist in say job
interviews or loan applications get
amplified by algorithms or robotic
soldiers behave unethically in the
battlefield
these are very legitimate concerns and
they deserve our attention as is the
case with any disruptive technology we
must ensure fair and ethical
applications of AI and I think we can
but the second kind of worry is deeper
it’s visceral it’s existential it’s the
worry that we people won’t be that
special anymore that Westworld got it
right we won’t be able to tell that
people from the machines there’s a
worst-case version of the worry in which
the machines take over and replace us as
the dominant species you’ve seen the
salaat in science fiction and also in
recent popular books you know the
narrative you know we humans are made
our contribution to evolution thank you
very much but our service is no longer
needed
but even ignoring this evolutionary
threat which by the way I think is
overblown the question remains is there
something that separates us from
machines I’m going to answer this in two
steps first I’ll scare the hell out of
you
and then I’ll not only calm you down but
hopefully even excite you about the
future
so first the scary part and I’ll say it
right now I believe that computers will
exist exhibit most if not all of the
traits that we view as uniquely human
and yes that includes thinking and
understanding and freewill and all the
rest I’m not alone in thinking this most
computer scientists I know tend to think
this way it includes for example Alan
Turing perhaps the most brilliant
computer science of all times among
other things
Turing invented what we today call the
Turing test which is a practical test of
whether computer is intelligent and
Turing thoughts already in 1950 that
there was no reason computers could not
and it includes John McCarthy the father
of AI who coined the term in the 1950s
now something this isn’t real thinking
or understanding the thing is I’ve yet
to see a single good argument perhaps
the best argument I’ve seen was given by
Ada Lovelace what makes her argument
striking by the way is that
in 1842 a century before Turing she was
a very talented writer mathematician
society woman she happened to be also
the daughter of Lord Byron and she’s
best known today for her work with
Charles Babbage Babbage was a also a
mathematician and he invented the first
general-purpose computer although I
never built it called the analytical
engine he notes that lovely wrote about
the analytical engine she included what
many considered to be the first computer
program in the world and even more
remarkably she speculated about what
today we would call artificial
intelligence specifically she argued
against the possibility of the computer
very sensible the computer just follows
instructions it has no original thought
think about that this is 1842 a century
before the first electronic computer
it’s mind-blowing its mind-blowing but
it’s also flawed in two ways first while
Lovelace ISM insight into computing was
astounding for her time by today’s
standards this naive today AI is not
characterized by handcrafted logical
rules but by machine learning operating
on machines a million times more
powerful than she could imagine giving
rise to programs that no human could
have explicitly coded or even
anticipated whether these are programs
that beat a world champion of the game
of Go or paint fantastical paintings the
second flaw is even more basic notice
let’s look again at what she wrote
notice that she doesn’t give an argument
she dresses her belief and appeals to
our intuition and that is the case with
all the arguments that followed the OL
point boiled down to the idea of command
this is ridiculous it just seems obvious
to the critics
that the computer can never really
understand or think or what have you it
can only mimic it superficially really
and how would you know the real thing if
you saw it it reminds me of one of the
stories of the Wiseman of helm that
fictional group are very silly people
from Eastern European Jewish folklore in
this story the man gets up in the
morning and says goodbye to his wife and
kids and sets out from the small town of
helm to the big town city of Warsaw and
along the way gets lost through some
funny circumstances and ends back in
hell but thinking arrived in Warsaw and
he’s struck by its resemblance to his
hometown there’s a street just like his
and a house very much like he’s an
amazingly a family so much like his own
family nice people they invite him in
they feed him and he wants when he wants
to leave they won’t let him so we spent
the recipe line is life with these
wonderful people and is always homesick
and that is how it is with the sceptics
of AI it doesn’t matter if the computer
works like a duck and talks like a DAC
for them somehow the real thing is
mysteriously different now am I saying
that I know for a fact that a computer
will exhibit the full range of human
mental activity no I’m quite sure about
thinking and understanding if you just
look at even today’s programs that
diagnose diseases better than doctors
that approve loan applications much more
reasonably than humans that grade
homework online and so on how good must
they be before you concede that they
understand what they’re talking about it
may not be exactly our understanding but
we don’t say that an airplane can’t fly
just because it doesn’t fly like a bird
and I’m also quite sure about creativity
and it’s not only because of that
painting we saw earlier on a recent
version of Google Translate the program
that translates between languages like
English and French invented or on its
own and internal language to help with
this translation it surprised everyone
not the least the people who wrote the
program
and I’m told that expert go programs
were in awe of the originality of the
alphago program so much so that they
named one of the moves after the program
so I’m a YES on thinking understanding
creativity I tend to think that the same
is true of things like feelings and
freewill and consciousness although I’m
not yet ready to argue it these are more
subtle notions certainly some have
argued like the late John McCarthy who I
mentioned earlier on I myself think that
the jury is still out but I wouldn’t bet
against the machine so this is scary
computers can think understand be
creative and maybe even have feelings
and free will and be conscious that mean
that we’re nothing but glorified laptops
perhaps to be replaced by a newer and
better model it doesn’t feel very good
and so natural to reject an idea because
it scares us but fear is a very poor
guide and this is where the story gets
happy again if you just shift
perspective you see not a threat but a
very exciting opportunity you see that
it’s not about the answers it’s about
the questions and it’s not about
computers versus humans then versus us
it’s not even about them at all it’s
really about us and that’s for two
reasons first computers already now
extend our cognitive skills but they
tend to be physically separate and made
of metal but that’ll change our nervous
system will connect wirelessly to giant
data centers and our blood streams will
carry thousands of nano computers
fighting diseases so it’ll be less clear
where we end in the computer begins but
even more basically we just don’t
understand these notions take free will
I’m sure that when you bear you voted
earlier on on on free will you were
asking yourself well what is it and what
does it mean for me to have free will
that’s certainly what the students in my
class asked and they quickly concluded
that we don’t know certainly a ton has
been written about it I recently spoke
with a very smart friend of mine a
technologist who happen to be religious
and I asked him about the famous quote
of Rabbi Akiva I called Sephora Bella
shoot mucuna all he’s foreseen and free
will is granted I asked him whether he
needed in his opinion it resolved the
obvious tension between divine
omniscience and free will he admitted
that probably not although later he sent
me some writing by his rabbi on it
philosophers of course had done battle
with free will for a long time for
example there’s a wonderful story by the
philosopher and logician raiments
mullion the story is not really this is
not Dumbledore this is mullion although
it’s true the smullin was also a
magician the story is called is God a
Taoists and the moral of the story is
that it makes no logical sense for a
person to not want to have free will
think about it’s fascinating perhaps my
favorite quote on free will comes from
literature from Isaac to share the
zingers Nobel Prize speech where we said
we have to believe in free will we have
no choice what all these quotes have in
common is that they’re clever they’re
thought-provoking they’re entertaining
they don’t tell us what free will is and
we’re in no better shape when it comes
to consciousness and perhaps only
marginally better when it comes to
feelings so it’s not that we don’t
understand computers we don’t understand
ourselves and this to me is where it
gets exciting
you see AI in addition to its practical
contributions is propelling us towards
the ultimate intellectual frontier
understanding our minds our souls
ourselves because at the end what do we
care about we certainly care about not
dying of cold and hunger and not being
eaten by predators and that’s also true
of lizards we want to love and be loved
and that’s true of my dog and then what
we have the sense that is more to our
life than survival and companionship we
don’t end the day and we say I wasn’t
eaten by a lion that was a good day we
look back at the day we look back in our
life and we take satisfaction in what
you
in what we’ve learned on how we acted on
the impact we had on others are the nice
things people said about us but for that
to make sense we have to think of
ourselves as creatures with thoughts and
beliefs and feelings aware of ourselves
and of each other
able to take our actions freely so to
understand ourselves we need to
understand these things
we have various tools for this we have
neuroscience and yes we have philosophy
of the rest but AI is giving us the
really interesting new lens through
which to think about these issues the
question about whether the question of
whether machines can embody these
notions and we should ask this question
boldly perhaps we’ll find that essential
nugget that separates us from machines
and certainly the contrast with machines
will tell us something about ourselves
perhaps we won’t find that nugget that
will also tell us something about
ourselves and it won’t demean us either
our sense of self-worth shouldn’t depend
on putting down the machine in both
cases we learn something about ourselves
and as the great philosopher Martha
Stewart said that’s a good thing before
I end I have a request please all go to
the website could computers o RG and
vote on these six questions this goes to
everyone here in the audience or
watching it live elsewhere or recording
later on please go ahead and vote and
you can vote many times perhaps you’ll
read something interesting the New York
Times or a la ohm maybe you’ll have a
conversation with your yoga instructor
maybe you’ll hear an interesting TED
talk by all means go ahead and revote by
the way they should have allowed that
for brexit but that’s a different story
we will collect these votes and see how
as a society we feel about these issues
and how that evolved over time and let’s
not only vote let’s start a conversation
about it and if we’re lucky we will have
a better understanding
not only of computers and of AI but also
of what it means to be a human being thank