-
Not Synced
Okay, so, good morning everyone.
-
Not Synced
I'll just get started.
-
Not Synced
My name is Shailesh and I give these talks
-
Not Synced
almost every year so this is a very deja-vu feeling for me.
-
Not Synced
The only thing different this time
-
Not Synced
is the stage is slightly thinner.
-
Not Synced
But great crowd, great list of talks so far.
-
Not Synced
So, Daniel called me a couple of weeks ago and said
-
Not Synced
"Why don't you give a keynote again?"
-
Not Synced
And I said, "You know, I'm running out of things to say now."
-
Not Synced
I've given four talks at different forums
-
Not Synced
with The Fifth Elephant and I wasn't so sure
-
Not Synced
what I want to talk about.
-
Not Synced
So, then, one of these days I was talking
-
Not Synced
to one of my non-geek friends
-
Not Synced
and he was very excited about what I do
-
Not Synced
so he said, 'What do you do?'
-
Not Synced
and I, you know, it was on the phone
-
Not Synced
and I started talking to him about this, that, and the other.
-
Not Synced
And for about 45 minutes I was rambling
-
Not Synced
and this guy was very quiet.
-
Not Synced
I didn't realise he wasn't a techie
-
Not Synced
and I was going on and on and after 45 minutes I stopped
-
Not Synced
and said, "Are you still there?
-
Not Synced
"Are you listening?"
-
Not Synced
And he said, "Yeah, I'm listening.
-
Not Synced
"Can you tell me what do you do again?"
(audience laughs)
-
Not Synced
And then I realized, how do I summarize this in 2 words?
-
Not Synced
So then I told him, "Hey, I'm building thinking machines."
-
Not Synced
And that's when he said, "Why didn't you say that before?
-
Not Synced
"It was so easy to say that, right?"
-
Not Synced
So that's how the title came by
-
Not Synced
and obviously we're not building thinking machines
-
Not Synced
but what I'm going talk about is towards thinking machines, right?
-
Not Synced
So, we have a long way to go.
-
Not Synced
So I added the word "towards" later.
-
Not Synced
So what I'm gonna talk about is all over the place.
-
Not Synced
I'm going to talk about philosophy, science fiction.
-
Not Synced
I'm going to talk about algorithms
-
Not Synced
and I'm going to talk about, you know, deep learning
-
Not Synced
and how to think about things beyond deep learning.
-
Not Synced
All right?
-
Not Synced
And let me give you a perspective and then we'll start.
-
Not Synced
So I'll take questions at the end.
-
Not Synced
Start working this.
-
Not Synced
All right, so, I ended my last year's talk on this quotation
-
Not Synced
So I thought I'll start on this quotation this time.
-
Not Synced
So I like this quotation because it puts a lot
-
Not Synced
of things into perspective of what we're doing,
-
Not Synced
how our civilisation got here, and where we're headed.
-
Not Synced
So it says, "Our technology, our machines, is part of our humanity.
-
Not Synced
"We created them to extend ourselves
-
Not Synced
"and that is what is unique about human beings!"
-
Not Synced
And if you look at chairs, and dogs, and animals, and cats
-
Not Synced
they don't create machines to extend themselves.
-
Not Synced
They just have instincts and they follow their instincts.
-
Not Synced
Right, that's very unique about human civilization.
-
Not Synced
We've created Taj Mahal, and space flights, and internet
-
Not Synced
and so we've come a very long way.
-
Not Synced
So if you think about the tools, right?
-
Not Synced
The cavemen had tools and now we have
-
Not Synced
a completely robotic assembly line with no humans
-
Not Synced
and you could turn the lights off and nothing will happen
-
Not Synced
the cars will get produced, right?
-
Not Synced
We've gone from just on-road, bullock carts,
-
Not Synced
to massive amounts of transportation that we can do now.
-
Not Synced
If you look at our ability to look further
-
Not Synced
into space, again, since Galileo,
-
Not Synced
we have made a lot of progress.
-
Not Synced
Recently we saw the news of Pluto flyby
-
Not Synced
so now we're able to send satellites into space.
-
Not Synced
If you look at the first computer we built
-
Not Synced
and where we are today, right?
-
Not Synced
We have a huge data center, and really
-
Not Synced
if you look at the whole thing in perspective
-
Not Synced
we have made an enormous amount of progress
-
Not Synced
in the last so many centuries, right?
-
Not Synced
So if you look just at the technical part
-
Not Synced
the IT kind of intelligent machines
-
Not Synced
we're not talking about mixies
-
Not Synced
and other things, just look at what AI
-
Not Synced
and deep learning, this stuff, has produced.
-
Not Synced
Today's machines can play chess.
-
Not Synced
And there's no human on the planet
-
Not Synced
who can play chess better than the machine.
-
Not Synced
I want to take a pause and think about where we are.
-
Not Synced
There's no human on the planet
-
Not Synced
who can play chess better than the machine.
-
Not Synced
There's no human on the planet
-
Not Synced
who can play Jeopardy better than the machine.
-
Not Synced
And recently, Google came out with automatic cars
-
Not Synced
so the machines can drive cars and record show
-
Not Synced
that these cars are better than humans under ideal conditions
-
Not Synced
And they have much less accident rates
-
Not Synced
and all the accidents happened because of other humans drivers.
-
Not Synced
They're not because of cars.
-
Not Synced
And recently also saw how machines
-
Not Synced
are able to create pictures, right?
-
Not Synced
So this is one of the things
-
Not Synced
that deep learning is internally doing.
-
Not Synced
And now think about all this.
-
Not Synced
Just think about where machines have gone today.
-
Not Synced
How many things they can do
-
Not Synced
which are way beyond our imagination
-
Not Synced
that machines could have done.
-
Not Synced
So obviously there's a lot they've done.
-
Not Synced
But can they do the following?
-
Not Synced
So obviously there's a lot they've done.
-
Not Synced
So one of the holy grails of AI is to have a machine
-
Not Synced
have a conversation with a human being.
-
Not Synced
We all know the Turing test and the
-
Not Synced
repercussions of this will be huge.
-
Not Synced
We could think about how we talk to the internet today.
-
Not Synced
We carefully craft word-for-word queries, right?
-
Not Synced
And you know, we allow the internet to make mistakes.
-
Not Synced
We craft queries again, and we take the suggestions or not.
-
Not Synced
We talk to the internet like we're talking to a three-year-old.
-
Not Synced
Now in the daily needs of massive data computers, NLP
-
Not Synced
and all its deploying staff, imagine how shameful it is
-
Not Synced
to talk to a computer like a 3-year-old.
-
Not Synced
So it's got the capacity of thousands of people
-
Not Synced
but it can't understand language.
-
Not Synced
So we need to change that.
-
Not Synced
Now imagine beyond keywords what can happen.
-
Not Synced
We can do question answering
-
Not Synced
but how do we do question answering today?
-
Not Synced
We've created Yahoo Answers, we've created Quora
-
Not Synced
where people can type questions
-
Not Synced
We do a match between the questions and the answers
-
Not Synced
and then we again do retrieval.
-
Not Synced
So not answering questions.
-
Not Synced
Now think about conversations.
Conversation is an even more complex thing
-
Not Synced
If it works out, what are the
repercussions? I don't want to study
-
Not Synced
physics from my physics teacher. I want to
study from Einstein or Feynman.
-
Not Synced
We already know all the language and
knowledge of these people.
-
Not Synced
Can we not have a persona or a person
Feynman or Einstein,
-
Not Synced
and have a conversation with that person,
right? So just imagine the future
-
Not Synced
of what will happen if we're just able to
have conversations with the machines.
-
Not Synced
So there's a long way to go between
keyword search and conversations.
-
Not Synced
Can we discover a cure for cancer?
There are a lot of diseases out there.
-
Not Synced
Now obviously there is a lot of research
pharma companies are doing.
-
Not Synced
There's a lot of new initiatives on how
to use the high end machine learning
-
Not Synced
in pharma research. But my contention is
I believe that the cure for a lot of
-
Not Synced
diseases is already out there. In all the
medical literature, if somebody could
-
Not Synced
actually read them, hold that knowledge in
the brain, in RAM, and do interconnections
-
Not Synced
we should be able to find a lot of things.
But what is the problem?
-
Not Synced
A single human expert, even in one field
cannot keep up with that quest of
-
Not Synced
knowledge, right. We forget some things,
we want to read certain papers.
-
Not Synced
And therefore, it's the other problem.
We have too much knowledge
-
Not Synced
and our individual brains are not
capable of forming those connections
-
Not Synced
in the - because we can't even read that
many docs, right?
-
Not Synced
But machines could do it, the way, and
then there's progress.
-
Not Synced
Can we not find cures or new medicine
too.
-
Not Synced
Can I crack the next IIT Entrance Exam?
You laughing today, but you never know.
-
Not Synced
Five years from now, what will happen?
We should hope that if Watson is
-
Not Synced
a test of intelligence, if Igloo is a test
of intelligence, could this not be
-
Not Synced
a test of intelligence.
The ability of AI system to be able to
-
Not Synced
actually solve an IIT paper and get a
rank 1.
-
Not Synced
What about, can I search all the video
scenes, which only have a goal shot
-
Not Synced
in the football videos and nothing else.
I don't want to watch the rest of it.
-
Not Synced
A lot of balls going here and there.
I just want to see the goal shots.
-
Not Synced
Today I cannot do that.
Can my machines be intelligent enough
-
Not Synced
to vision part, to actually find, this is
a goal, this is a goal, this is a goal -
-
Not Synced
the rest of it is something else.
So we can imagine the applications now.
-
Not Synced
We were talking about sarcasm a lot and we
all understand sarcasm is a very hard
-
Not Synced
thing to do. And imagine if you could
detect sarcasm, what else can you do?
-
Not Synced
You writing an email to your boss
You're angry, you've written
-
Not Synced
a sarcastic comment, and ? says,
'Hey are you sure about this?'
-
Not Synced
In the heat of the moment, can
I put it this way?
-
Not Synced
So, like, today we do attachments. Can you
detect sarcasm and things like that.
-
Not Synced
And to me the holy grail of AI is not
really all these big things,
-
Not Synced
but a really simple thing. Can a machine
find a joke funny?
-
Not Synced
Now there are a lot of - don't know if you
guys watch Star Trek - but data entry
-
Not Synced
300, 400 years from now, is an android who
is capable of all these other things.
-
Not Synced
He's a great supercomputer in human form
but he's still struggling with humans.
-
Not Synced
That's how hard the problem is.
So obviously we have a long way to go.
-
Not Synced
We've come a long way and we have a long
way to go.
-
Not Synced
So this talk is really about the way
forward.
-
Not Synced
So, what do we imagine the future to be?
We want something like this.
-
Not Synced
Golden ?
We all want a Jarvis, right?
-
Not Synced
Who takes care of the chores and gets rid
of the whatever and then we all want
-
Not Synced
a Jarvis right? So if you watch these
movies again, after watching this talk,
-
Not Synced
you'll have a very different perspective
on what we need to do to get here.
-
Not Synced
It's not going to happen just because
we're going to make more and more
-
Not Synced
Hollywood movies like this.