[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:01.10,0:00:03.24,Default,,0000,0000,0000,,- Okay, so, good morning everyone.
Dialogue: 0,0:00:03.24,0:00:04.91,Default,,0000,0000,0000,,I'll just get started.
Dialogue: 0,0:00:04.91,0:00:08.84,Default,,0000,0000,0000,,My name is Shailesh and I give these talks
Dialogue: 0,0:00:08.93,0:00:11.95,Default,,0000,0000,0000,,almost every year so this is a very deja-vu feeling for me.
Dialogue: 0,0:00:11.99,0:00:13.33,Default,,0000,0000,0000,,The only thing different this time
Dialogue: 0,0:00:13.33,0:00:15.73,Default,,0000,0000,0000,,is the stage is slightly thinner.
Dialogue: 0,0:00:15.73,0:00:21.88,Default,,0000,0000,0000,,But great crowd, great list of talks so far.
Dialogue: 0,0:00:21.88,0:00:25.87,Default,,0000,0000,0000,,So, Daniel called me a couple of weeks ago and said
Dialogue: 0,0:00:25.87,0:00:27.70,Default,,0000,0000,0000,,"Why don't you give a keynote again?"
Dialogue: 0,0:00:27.70,0:00:30.69,Default,,0000,0000,0000,,And I said, "You know, I'm running out of things to say now."
Dialogue: 0,0:00:30.69,0:00:33.20,Default,,0000,0000,0000,,I've given four talks at different forums
Dialogue: 0,0:00:33.20,0:00:36.88,Default,,0000,0000,0000,,with The Fifth Elephant and I wasn't sure
Dialogue: 0,0:00:36.88,0:00:38.03,Default,,0000,0000,0000,,what I'm gonna talk about
Dialogue: 0,0:00:38.03,0:00:41.32,Default,,0000,0000,0000,,So, then, one of these days I was talking
Dialogue: 0,0:00:41.32,0:00:43.28,Default,,0000,0000,0000,,to one of my non-geek friends
Dialogue: 0,0:00:43.28,0:00:46.22,Default,,0000,0000,0000,,and he was very excited about what I do
Dialogue: 0,0:00:46.22,0:00:47.46,Default,,0000,0000,0000,,so he said, 'What do you do?'
Dialogue: 0,0:00:47.46,0:00:49.33,Default,,0000,0000,0000,,and I, you know, it was on the phone
Dialogue: 0,0:00:49.33,0:00:53.03,Default,,0000,0000,0000,,and I started talking to him about this, that, and the other.
Dialogue: 0,0:00:53.03,0:00:55.20,Default,,0000,0000,0000,,And for about 45 minutes I was rambling
Dialogue: 0,0:00:55.20,0:00:57.08,Default,,0000,0000,0000,,and this guy was very quiet.
Dialogue: 0,0:00:57.08,0:01:01.89,Default,,0000,0000,0000,,I didn't realize he wasn't a techie
Dialogue: 0,0:01:01.89,0:01:06.76,Default,,0000,0000,0000,,and I was going on and on and after 45 minutes I stopped
Dialogue: 0,0:01:06.76,0:01:09.18,Default,,0000,0000,0000,,and said, "Are you still there? Are you listening?"
Dialogue: 0,0:01:09.18,0:01:11.12,Default,,0000,0000,0000,,And he said, "Yeah, I'm listening.
Dialogue: 0,0:01:12.02,0:01:15.78,Default,,0000,0000,0000,,"Can you tell me what do you do again?"\N(audience laughs)
Dialogue: 0,0:01:15.78,0:01:20.28,Default,,0000,0000,0000,,And then I realized, how do I summarize this in two words?
Dialogue: 0,0:01:20.28,0:01:24.19,Default,,0000,0000,0000,,So then I told him, "Hey, I'm building thinking machines."
Dialogue: 0,0:01:24.19,0:01:26.66,Default,,0000,0000,0000,,And that's when he said, "Why didn't you say that before?
Dialogue: 0,0:01:26.66,0:01:28.42,Default,,0000,0000,0000,,"It was so easy to say that, right?"
Dialogue: 0,0:01:28.42,0:01:30.13,Default,,0000,0000,0000,,So that's how the title came by
Dialogue: 0,0:01:30.13,0:01:33.52,Default,,0000,0000,0000,,and obviously we're not building thinking machines
Dialogue: 0,0:01:33.52,0:01:36.59,Default,,0000,0000,0000,,but what I'm gonna talk about is towards thinking machines, right?
Dialogue: 0,0:01:36.59,0:01:38.95,Default,,0000,0000,0000,,So, we have a long way to go.
Dialogue: 0,0:01:38.95,0:01:42.33,Default,,0000,0000,0000,,So I added the word "towards" later.
Dialogue: 0,0:01:42.33,0:01:45.82,Default,,0000,0000,0000,,So what I'm gonna talk about is all over the place.
Dialogue: 0,0:01:45.82,0:01:48.70,Default,,0000,0000,0000,,I'm gonna talk about philosophy, science fiction.
Dialogue: 0,0:01:48.70,0:01:51.44,Default,,0000,0000,0000,,I'll talk about algorithms
Dialogue: 0,0:01:51.44,0:01:54.66,Default,,0000,0000,0000,,and I'm gonna talk about deep learning
Dialogue: 0,0:01:54.66,0:01:57.88,Default,,0000,0000,0000,,and how to think about things beyond deep learning.
Dialogue: 0,0:01:57.88,0:02:02.42,Default,,0000,0000,0000,,And let me give you a perspective and then we'll start.
Dialogue: 0,0:02:02.42,0:02:04.58,Default,,0000,0000,0000,,So I'll take questions at the end.
Dialogue: 0,0:02:04.58,0:02:05.66,Default,,0000,0000,0000,,It's not working.
Dialogue: 0,0:02:09.48,0:02:11.02,Default,,0000,0000,0000,,It's not working, this.
Dialogue: 0,0:02:39.38,0:02:40.84,Default,,0000,0000,0000,,That's fine.
Dialogue: 0,0:02:40.84,0:02:45.33,Default,,0000,0000,0000,,All right, so, I ended my last year's talk on this quotation
Dialogue: 0,0:02:45.33,0:02:47.85,Default,,0000,0000,0000,,So I thought I'll start on this quotation this time.
Dialogue: 0,0:02:48.60,0:02:50.58,Default,,0000,0000,0000,,So I like this quotation because it puts a lot
Dialogue: 0,0:02:50.58,0:02:53.72,Default,,0000,0000,0000,,of things into perspective of what we're doing,
Dialogue: 0,0:02:53.72,0:02:56.06,Default,,0000,0000,0000,,how our civilization got here
Dialogue: 0,0:02:56.06,0:02:57.49,Default,,0000,0000,0000,,and where we are headed.
Dialogue: 0,0:02:58.23,0:03:02.89,Default,,0000,0000,0000,,So it says, "Our technology, our machines, is part of our humanity.
Dialogue: 0,0:03:02.89,0:03:05.92,Default,,0000,0000,0000,,"We created them to extend ourselves
Dialogue: 0,0:03:05.92,0:03:08.13,Default,,0000,0000,0000,,"and that is what is unique about human beings!"
Dialogue: 0,0:03:08.13,0:03:11.52,Default,,0000,0000,0000,,And if you look at chairs, and dogs, and animals, and cats
Dialogue: 0,0:03:11.52,0:03:15.13,Default,,0000,0000,0000,,they don't create machines to extend themselves.
Dialogue: 0,0:03:15.13,0:03:17.33,Default,,0000,0000,0000,,They just have instincts and they follow their instincts.
Dialogue: 0,0:03:17.33,0:03:20.02,Default,,0000,0000,0000,,Right, that's very unique about human civilization.
Dialogue: 0,0:03:20.02,0:03:24.62,Default,,0000,0000,0000,,We've created Taj Mahal, and space flights, and internet.
Dialogue: 0,0:03:24.62,0:03:26.95,Default,,0000,0000,0000,,So we've come a very long way.
Dialogue: 0,0:03:26.95,0:03:29.12,Default,,0000,0000,0000,,So if you think about the tools, right?
Dialogue: 0,0:03:29.12,0:03:33.25,Default,,0000,0000,0000,,The cavemen had tools and now we have
Dialogue: 0,0:03:33.25,0:03:37.89,Default,,0000,0000,0000,,a completely robotic assembly line with no humans
Dialogue: 0,0:03:37.89,0:03:40.22,Default,,0000,0000,0000,,and you could turn the lights off and nothing will happen
Dialogue: 0,0:03:40.22,0:03:41.100,Default,,0000,0000,0000,,the cars will get produced, right?
Dialogue: 0,0:03:41.100,0:03:44.21,Default,,0000,0000,0000,,If you look at our transportation
Dialogue: 0,0:03:44.98,0:03:50.08,Default,,0000,0000,0000,,we have gone from just on-road, bullock carts,
Dialogue: 0,0:03:50.08,0:03:54.18,Default,,0000,0000,0000,,to massive amounts of transportation that we can do now.
Dialogue: 0,0:03:56.48,0:03:58.98,Default,,0000,0000,0000,,If you look at our ability to look further
Dialogue: 0,0:03:58.98,0:04:02.17,Default,,0000,0000,0000,,into space, again...
Dialogue: 0,0:04:06.48,0:04:08.68,Default,,0000,0000,0000,,Since Galileo, we have made a lot of progress.
Dialogue: 0,0:04:08.68,0:04:12.21,Default,,0000,0000,0000,,Recently we saw the news of Pluto flyby
Dialogue: 0,0:04:12.21,0:04:14.85,Default,,0000,0000,0000,,so now we're able to send satellites into space.
Dialogue: 0,0:04:14.85,0:04:17.72,Default,,0000,0000,0000,,If you look at the first computer we built
Dialogue: 0,0:04:17.72,0:04:19.30,Default,,0000,0000,0000,,and where we are today, right?
Dialogue: 0,0:04:19.30,0:04:22.21,Default,,0000,0000,0000,,We have a huge data center, and really
Dialogue: 0,0:04:22.21,0:04:24.16,Default,,0000,0000,0000,,if you look at the whole thing in perspective
Dialogue: 0,0:04:24.16,0:04:26.50,Default,,0000,0000,0000,,we have made an enormous amount of progress
Dialogue: 0,0:04:26.50,0:04:29.87,Default,,0000,0000,0000,,in the last so many centuries, right?
Dialogue: 0,0:04:29.87,0:04:32.58,Default,,0000,0000,0000,,So if you look just at the technical part
Dialogue: 0,0:04:32.58,0:04:35.30,Default,,0000,0000,0000,,the IT kind of intelligent machines
Dialogue: 0,0:04:35.30,0:04:36.67,Default,,0000,0000,0000,,we're not talking about mixies
Dialogue: 0,0:04:36.67,0:04:40.46,Default,,0000,0000,0000,,and other things, just look at what AI
Dialogue: 0,0:04:40.46,0:04:43.62,Default,,0000,0000,0000,,and deep learning and all this stuff has produced.
Dialogue: 0,0:04:43.62,0:04:45.90,Default,,0000,0000,0000,,Today's machines can play chess.
Dialogue: 0,0:04:45.90,0:04:47.75,Default,,0000,0000,0000,,And there's no human on the planet
Dialogue: 0,0:04:47.75,0:04:50.41,Default,,0000,0000,0000,,who can play chess better than the machine.
Dialogue: 0,0:04:50.41,0:04:54.71,Default,,0000,0000,0000,,I want to take a pause and think about where we are.
Dialogue: 0,0:04:54.71,0:04:56.74,Default,,0000,0000,0000,,There's no human on the planet
Dialogue: 0,0:04:56.74,0:04:58.88,Default,,0000,0000,0000,,who can play chess better than a machine.
Dialogue: 0,0:05:00.24,0:05:02.60,Default,,0000,0000,0000,,There's no human on the planet
Dialogue: 0,0:05:02.60,0:05:05.02,Default,,0000,0000,0000,,who can play Jeopardy better than a machine.
Dialogue: 0,0:05:06.96,0:05:10.97,Default,,0000,0000,0000,,And recently, Google came out with automatic cars
Dialogue: 0,0:05:10.97,0:05:13.84,Default,,0000,0000,0000,,so the machines can drive cars and record show
Dialogue: 0,0:05:13.84,0:05:18.90,Default,,0000,0000,0000,,that these cars are better than humans under ideal conditions
Dialogue: 0,0:05:18.90,0:05:21.44,Default,,0000,0000,0000,,And they have much less accident rates
Dialogue: 0,0:05:21.44,0:05:25.03,Default,,0000,0000,0000,,and all the accidents happened because of other humans drivers.
Dialogue: 0,0:05:25.03,0:05:26.35,Default,,0000,0000,0000,,They're not because of cars.
Dialogue: 0,0:05:27.91,0:05:29.100,Default,,0000,0000,0000,,And recently you also saw
Dialogue: 0,0:05:29.100,0:05:32.73,Default,,0000,0000,0000,,how machines are able to create pictures, right?
Dialogue: 0,0:05:32.73,0:05:34.42,Default,,0000,0000,0000,,So this is one of the things
Dialogue: 0,0:05:34.42,0:05:37.75,Default,,0000,0000,0000,,that we saw what deep learning is internally doing.
Dialogue: 0,0:05:37.75,0:05:39.30,Default,,0000,0000,0000,,And now think about all this.
Dialogue: 0,0:05:39.30,0:05:44.06,Default,,0000,0000,0000,,Just think about where machines have gone today.
Dialogue: 0,0:05:44.06,0:05:45.86,Default,,0000,0000,0000,,How many things they can do
Dialogue: 0,0:05:45.86,0:05:48.26,Default,,0000,0000,0000,,which are way beyond our imagination
Dialogue: 0,0:05:48.26,0:05:49.76,Default,,0000,0000,0000,,that machines could have done.
Dialogue: 0,0:05:51.14,0:05:55.04,Default,,0000,0000,0000,,So obviously there's a lot they've done.
Dialogue: 0,0:05:55.04,0:05:57.18,Default,,0000,0000,0000,,But can they do the following?
Dialogue: 0,0:05:57.18,0:05:59.24,Default,,0000,0000,0000,,We would want to stress the limits
Dialogue: 0,0:05:59.24,0:06:02.40,Default,,0000,0000,0000,,so one of the holy grails of AI
Dialogue: 0,0:06:02.40,0:06:06.03,Default,,0000,0000,0000,,is to have a machine have a conversation with a human being.
Dialogue: 0,0:06:06.03,0:06:07.86,Default,,0000,0000,0000,,We all know the Turing test
Dialogue: 0,0:06:07.86,0:06:11.11,Default,,0000,0000,0000,,and the repercussions of this will be huge.
Dialogue: 0,0:06:11.11,0:06:14.47,Default,,0000,0000,0000,,If you think about how we talk to the internet today
Dialogue: 0,0:06:14.47,0:06:18.97,Default,,0000,0000,0000,,we carefully craft three-word, four-word queries, right?
Dialogue: 0,0:06:18.97,0:06:21.95,Default,,0000,0000,0000,,And you know, we allow the internet to make mistakes.
Dialogue: 0,0:06:21.95,0:06:25.11,Default,,0000,0000,0000,,We craft the queries again, we take the suggestions or not.
Dialogue: 0,0:06:25.11,0:06:28.64,Default,,0000,0000,0000,,We talk to the internet like we're talking to a three-year-old.
Dialogue: 0,0:06:28.64,0:06:33.20,Default,,0000,0000,0000,,Now in the day and age needs of massive data computers, NLP
Dialogue: 0,0:06:33.20,0:06:37.21,Default,,0000,0000,0000,,and all this deep-learning stuff, imagine what a shameful thing it is
Dialogue: 0,0:06:37.21,0:06:40.09,Default,,0000,0000,0000,,to talk to a computer like a 3-year-old.
Dialogue: 0,0:06:40.09,0:06:43.99,Default,,0000,0000,0000,,So it's got the capacity of thousands of people
Dialogue: 0,0:06:43.99,0:06:46.40,Default,,0000,0000,0000,,but it can't understand language.
Dialogue: 0,0:06:46.40,0:06:48.20,Default,,0000,0000,0000,,So we need to change that.
Dialogue: 0,0:06:48.20,0:06:50.65,Default,,0000,0000,0000,,Now imagine beyond keywords what can happen.
Dialogue: 0,0:06:50.65,0:06:52.25,Default,,0000,0000,0000,,We can do question-answering
Dialogue: 0,0:06:52.25,0:06:54.63,Default,,0000,0000,0000,,but how do we do question-answering today?
Dialogue: 0,0:06:54.63,0:06:57.64,Default,,0000,0000,0000,,We have created Yahoo Answers, we have created Quora
Dialogue: 0,0:06:57.64,0:07:01.14,Default,,0000,0000,0000,,and people who type questions, we do a match.
Dialogue: 0,0:07:01.14,0:07:04.88,Default,,0000,0000,0000,,Between the questions and the answers
Dialogue: 0,0:07:04.88,0:07:07.60,Default,,0000,0000,0000,,and then we again do retrieval.
Dialogue: 0,0:07:07.60,0:07:09.64,Default,,0000,0000,0000,,We're still not answering questions.
Dialogue: 0,0:07:10.52,0:07:11.98,Default,,0000,0000,0000,,Now think about conversations.
Dialogue: 0,0:07:11.98,0:07:14.16,Default,,0000,0000,0000,,Conversation is an even more complex thing.
Dialogue: 0,0:07:14.16,0:07:17.53,Default,,0000,0000,0000,,If it works out, what are the repercussions?
Dialogue: 0,0:07:17.53,0:07:20.16,Default,,0000,0000,0000,,I don't want to study physics from my physics teacher.
Dialogue: 0,0:07:20.16,0:07:22.84,Default,,0000,0000,0000,,I want to study it from Einstein or Feynman.
Dialogue: 0,0:07:23.71,0:07:28.26,Default,,0000,0000,0000,,We already know all the language and the knowledge of these people.
Dialogue: 0,0:07:28.26,0:07:32.02,Default,,0000,0000,0000,,Can we not have a persona of a person, Feynman or Einstein
Dialogue: 0,0:07:32.02,0:07:34.78,Default,,0000,0000,0000,,and have a conversation with that person?
Dialogue: 0,0:07:35.54,0:07:38.52,Default,,0000,0000,0000,,So, just imagine the future of what will happen
Dialogue: 0,0:07:38.52,0:07:41.55,Default,,0000,0000,0000,,if we are able to just have conversations with the machines.
Dialogue: 0,0:07:41.55,0:07:43.59,Default,,0000,0000,0000,,So, there's a long way to go between
Dialogue: 0,0:07:43.59,0:07:46.68,Default,,0000,0000,0000,,keyword search and conversations.
Dialogue: 0,0:07:46.68,0:07:48.99,Default,,0000,0000,0000,,Can we discover a cure for cancer?
Dialogue: 0,0:07:48.99,0:07:51.00,Default,,0000,0000,0000,,There are a lot of diseases out there.
Dialogue: 0,0:07:51.00,0:07:52.82,Default,,0000,0000,0000,,Now, obviously there is a lot
Dialogue: 0,0:07:52.82,0:07:54.84,Default,,0000,0000,0000,,of research pharma companies are doing.
Dialogue: 0,0:07:54.84,0:07:56.94,Default,,0000,0000,0000,,There's a lot of new initiatives in how
Dialogue: 0,0:07:56.94,0:08:00.16,Default,,0000,0000,0000,,to use the high-end machine learning in pharma research.
Dialogue: 0,0:08:00.99,0:08:06.14,Default,,0000,0000,0000,,But my contention is that I believe that the cure for a lot
Dialogue: 0,0:08:06.14,0:08:08.65,Default,,0000,0000,0000,,of diseases is already out there.
Dialogue: 0,0:08:08.65,0:08:11.47,Default,,0000,0000,0000,,In all the medical literature, if somebody
Dialogue: 0,0:08:11.47,0:08:14.45,Default,,0000,0000,0000,,could actually read them, hold that knowledge
Dialogue: 0,0:08:14.45,0:08:19.13,Default,,0000,0000,0000,,in the brain, in RAM, and do interconnections
Dialogue: 0,0:08:19.13,0:08:21.74,Default,,0000,0000,0000,,we should be able to find a lot of things.
Dialogue: 0,0:08:21.74,0:08:22.86,Default,,0000,0000,0000,,But what is the problem?
Dialogue: 0,0:08:22.86,0:08:25.92,Default,,0000,0000,0000,,A single human expert, even in one field
Dialogue: 0,0:08:25.92,0:08:28.73,Default,,0000,0000,0000,,cannot keep up with that quest of knowledge, right?
Dialogue: 0,0:08:28.73,0:08:31.72,Default,,0000,0000,0000,,We'll forget some things, we won't read certain papers.
Dialogue: 0,0:08:31.72,0:08:34.06,Default,,0000,0000,0000,,And therefore, it's the other problem.
Dialogue: 0,0:08:34.06,0:08:37.19,Default,,0000,0000,0000,,We have too much knowledge and our individual brains
Dialogue: 0,0:08:37.19,0:08:41.38,Default,,0000,0000,0000,,are not capable of forming those connections in the...
Dialogue: 0,0:08:41.38,0:08:44.20,Default,,0000,0000,0000,,Because we can't even read that many documents, right?
Dialogue: 0,0:08:44.20,0:08:46.04,Default,,0000,0000,0000,,But if a machine could do it
Dialogue: 0,0:08:46.04,0:08:48.05,Default,,0000,0000,0000,,the way NLP has progressed
Dialogue: 0,0:08:48.05,0:08:50.69,Default,,0000,0000,0000,,can we not find cures or new medicine?
Dialogue: 0,0:08:52.60,0:08:54.94,Default,,0000,0000,0000,,Can I crack the next IIT Entrance Exam?
Dialogue: 0,0:08:58.14,0:09:01.56,Default,,0000,0000,0000,,You're laughing today, but you never know.
Dialogue: 0,0:09:01.56,0:09:03.78,Default,,0000,0000,0000,,Five years from now, what will happen?
Dialogue: 0,0:09:03.78,0:09:08.74,Default,,0000,0000,0000,,And we should hope that if Watson is a test of intelligence
Dialogue: 0,0:09:08.74,0:09:11.55,Default,,0000,0000,0000,,if Igloo is a test of intelligence
Dialogue: 0,0:09:11.55,0:09:14.19,Default,,0000,0000,0000,,could this not be a test of intelligence?
Dialogue: 0,0:09:14.19,0:09:17.56,Default,,0000,0000,0000,,The ability of AI system to be able
Dialogue: 0,0:09:17.56,0:09:20.97,Default,,0000,0000,0000,,to actually solve an IIT paper and get a rank 1.
Dialogue: 0,0:09:23.61,0:09:26.99,Default,,0000,0000,0000,,Can I search all the video scenes
Dialogue: 0,0:09:26.99,0:09:28.61,Default,,0000,0000,0000,,which only have a goal shot
Dialogue: 0,0:09:28.61,0:09:31.60,Default,,0000,0000,0000,,in the football videos and nothing else.
Dialogue: 0,0:09:31.60,0:09:34.22,Default,,0000,0000,0000,,I don't want to watch the rest of it.
Dialogue: 0,0:09:34.22,0:09:36.04,Default,,0000,0000,0000,,A lot of balls going here and there.
Dialogue: 0,0:09:36.04,0:09:37.85,Default,,0000,0000,0000,,I just wanna see the goal shots.
Dialogue: 0,0:09:37.85,0:09:39.96,Default,,0000,0000,0000,,Today I cannot do that.
Dialogue: 0,0:09:39.96,0:09:42.14,Default,,0000,0000,0000,,Can my machines be intelligent enough
Dialogue: 0,0:09:42.14,0:09:44.43,Default,,0000,0000,0000,,the vision part, that can actually find
Dialogue: 0,0:09:44.43,0:09:46.33,Default,,0000,0000,0000,,this is a goal, this is a goal, this is a goal
Dialogue: 0,0:09:46.33,0:09:48.56,Default,,0000,0000,0000,,the rest of it is something else.
Dialogue: 0,0:09:48.56,0:09:50.92,Default,,0000,0000,0000,,So we can imagine the applications out there.
Dialogue: 0,0:09:51.82,0:09:54.02,Default,,0000,0000,0000,,We were talking about sarcasm a lot
Dialogue: 0,0:09:54.02,0:09:57.37,Default,,0000,0000,0000,,and we all understand sarcasm is a very hard thing to do.
Dialogue: 0,0:09:57.37,0:10:01.54,Default,,0000,0000,0000,,And imagine if you could detect sarcasm, what else can you do?
Dialogue: 0,0:10:01.54,0:10:03.44,Default,,0000,0000,0000,,You're writing an email to your boss
Dialogue: 0,0:10:03.44,0:10:06.73,Default,,0000,0000,0000,,you're angry, you have written a sarcastic comment
Dialogue: 0,0:10:06.73,0:10:09.26,Default,,0000,0000,0000,,and Gmail says, "Hey, are you sure about this?"
Dialogue: 0,0:10:09.26,0:10:12.09,Default,,0000,0000,0000,,In the heat of the moment\N(audience laughs)
Dialogue: 0,0:10:13.00,0:10:14.58,Default,,0000,0000,0000,,can I put it this way?
Dialogue: 0,0:10:15.52,0:10:17.58,Default,,0000,0000,0000,,So, like, today we do attachments
Dialogue: 0,0:10:17.58,0:10:19.82,Default,,0000,0000,0000,,can we detect sarcasm and things like that?
Dialogue: 0,0:10:20.73,0:10:25.50,Default,,0000,0000,0000,,And to me the holy grail of AI is not really
Dialogue: 0,0:10:25.50,0:10:28.26,Default,,0000,0000,0000,,all these big things, but a very simple thing.
Dialogue: 0,0:10:28.26,0:10:31.08,Default,,0000,0000,0000,,Can a machine find a joke funny?
Dialogue: 0,0:10:32.21,0:10:33.88,Default,,0000,0000,0000,,Now there are a lot of...
Dialogue: 0,0:10:33.88,0:10:36.01,Default,,0000,0000,0000,,I don't know if you guys watch Star Trek
Dialogue: 0,0:10:36.01,0:10:39.85,Default,,0000,0000,0000,,but Data, in 300 years, 400 years from now
Dialogue: 0,0:10:39.85,0:10:40.99,Default,,0000,0000,0000,,is an android.
Dialogue: 0,0:10:40.99,0:10:43.28,Default,,0000,0000,0000,,He is capable of all these other things.
Dialogue: 0,0:10:43.28,0:10:46.84,Default,,0000,0000,0000,,He's a great supercomputer in a human form
Dialogue: 0,0:10:46.84,0:10:48.70,Default,,0000,0000,0000,,but he's still struggling with humor.
Dialogue: 0,0:10:48.70,0:10:50.62,Default,,0000,0000,0000,,That's how hard the problem is.
Dialogue: 0,0:10:51.56,0:10:54.24,Default,,0000,0000,0000,,So obviously we have a long way to go.
Dialogue: 0,0:10:54.24,0:10:56.38,Default,,0000,0000,0000,,We have come a long way and we have a long way to go.
Dialogue: 0,0:10:56.38,0:10:59.29,Default,,0000,0000,0000,,So this talk is really about the way forward.
Dialogue: 0,0:11:00.96,0:11:04.20,Default,,0000,0000,0000,,So, what do we imagine the future to be?
Dialogue: 0,0:11:04.20,0:11:06.34,Default,,0000,0000,0000,,We want something like this.
Dialogue: 0,0:11:07.23,0:11:09.35,Default,,0000,0000,0000,,Good and bad, hopefully good.
Dialogue: 0,0:11:09.35,0:11:10.38,Default,,0000,0000,0000,,We want a Jarvis, right?
Dialogue: 0,0:11:10.38,0:11:11.74,Default,,0000,0000,0000,,We all want a Jarvis
Dialogue: 0,0:11:11.74,0:11:13.36,Default,,0000,0000,0000,,who'll takes care of the chores
Dialogue: 0,0:11:13.36,0:11:16.87,Default,,0000,0000,0000,,and get rid of whatever
Dialogue: 0,0:11:16.87,0:11:19.35,Default,,0000,0000,0000,,and we all want a Jarvis right?
Dialogue: 0,0:11:19.35,0:11:22.53,Default,,0000,0000,0000,,So if you watch these movies again
Dialogue: 0,0:11:22.53,0:11:24.56,Default,,0000,0000,0000,,after watching this talk
Dialogue: 0,0:11:24.56,0:11:26.56,Default,,0000,0000,0000,,you'll have a very different perspective
Dialogue: 0,0:11:26.56,0:11:29.71,Default,,0000,0000,0000,,on what we need to do to get here.
Dialogue: 0,0:11:29.71,0:11:31.68,Default,,0000,0000,0000,,It's not gonna happen just because we're
Dialogue: 0,0:11:31.68,0:11:35.04,Default,,0000,0000,0000,,gonna make more and more \NHollywood movies like this.
Dialogue: 0,0:11:35.34,0:11:38.55,Default,,0000,0000,0000,,I mean, Asimov wrote \N"I, Robot" in the 70s
Dialogue: 0,0:11:38.70,0:11:40.53,Default,,0000,0000,0000,,and we're still not there.
Dialogue: 0,0:11:41.40,0:11:44.10,Default,,0000,0000,0000,,It's not gonna happen because we keep \Ndoing "data science"
Dialogue: 0,0:11:45.30,0:11:47.84,Default,,0000,0000,0000,,And that's one of the reasons I wanted\Nto do this talk
Dialogue: 0,0:11:47.84,0:11:49.53,Default,,0000,0000,0000,,'cause a lot of people keep
Dialogue: 0,0:11:50.00,0:11:52.27,Default,,0000,0000,0000,,thinking "data science is the end \Nof the world"
Dialogue: 0,0:11:52.27,0:11:54.02,Default,,0000,0000,0000,,but there's a lot more to data science
Dialogue: 0,0:11:54.02,0:11:57.08,Default,,0000,0000,0000,,and I want to see how we can go beyond\Ndata science
Dialogue: 0,0:11:57.48,0:11:59.00,Default,,0000,0000,0000,,- and this is not data science.
Dialogue: 0,0:11:59.01,0:12:00.56,Default,,0000,0000,0000,,This is artificial intelligence.
Dialogue: 0,0:12:00.56,0:12:02.50,Default,,0000,0000,0000,,Right? So I want to draw the distinction
Dialogue: 0,0:12:02.51,0:12:04.65,Default,,0000,0000,0000,,and say how we can move\Nbeyond data science
Dialogue: 0,0:12:05.06,0:12:06.33,Default,,0000,0000,0000,,- nothing wrong with it -
Dialogue: 0,0:12:06.52,0:12:08.28,Default,,0000,0000,0000,,but it's, it's a done deal.
Dialogue: 0,0:12:08.87,0:12:11.41,Default,,0000,0000,0000,,Right? We have software you can download,
Dialogue: 0,0:12:11.41,0:12:13.17,Default,,0000,0000,0000,,you can put up whatever you want,
Dialogue: 0,0:12:13.48,0:12:16.65,Default,,0000,0000,0000,,it's a done deal. Data science has been\Npackaged, already.
Dialogue: 0,0:12:17.52,0:12:20.22,Default,,0000,0000,0000,,Right? If you look at Microsoft Azure,
Dialogue: 0,0:12:20.22,0:12:22.28,Default,,0000,0000,0000,,or some of these other softwares, right?
Dialogue: 0,0:12:22.30,0:12:23.71,Default,,0000,0000,0000,,It has already been packaged
Dialogue: 0,0:12:24.05,0:12:25.96,Default,,0000,0000,0000,,All you have to do is download the right
Dialogue: 0,0:12:25.96,0:12:28.36,Default,,0000,0000,0000,,software, put your data\Nin the right format,
Dialogue: 0,0:12:28.36,0:12:32.88,Default,,0000,0000,0000,,and you're done. Right? So there's nothing\N"great" about data science anymore.
Dialogue: 0,0:12:33.84,0:12:36.29,Default,,0000,0000,0000,,Sorry about that, but, you know,
Dialogue: 0,0:12:36.53,0:12:39.30,Default,,0000,0000,0000,,we need to jolt ourselves out of this\Ncomfort zone, and say
Dialogue: 0,0:12:39.30,0:12:40.94,Default,,0000,0000,0000,,"okay, we are all data scientists"
Dialogue: 0,0:12:40.94,0:12:42.05,Default,,0000,0000,0000,,- that's not it, right?
Dialogue: 0,0:12:42.10,0:12:44.56,Default,,0000,0000,0000,,How do we get here?\NHow will data science get here?
Dialogue: 0,0:12:45.13,0:12:49.18,Default,,0000,0000,0000,,Alright. So, we'll get here by asking\Na lot of deeper questions.
Dialogue: 0,0:12:49.36,0:12:51.30,Default,,0000,0000,0000,,Right? Not the questions like
Dialogue: 0,0:12:51.78,0:12:54.96,Default,,0000,0000,0000,,"Why is this customer\Nreturning from Flipkart?", right? or
Dialogue: 0,0:12:54.96,0:12:58.77,Default,,0000,0000,0000,,"Who's -- what is the next product to\Nrecommend to somebody?", or
Dialogue: 0,0:12:59.10,0:13:02.02,Default,,0000,0000,0000,,"Which movie you're going to ask?"\NThese are not the questions
Dialogue: 0,0:13:02.46,0:13:04.64,Default,,0000,0000,0000,,that'll take us to the next stage. Right?
Dialogue: 0,0:13:04.64,0:13:07.15,Default,,0000,0000,0000,,So the question that'll take us\Nto the next stage is
Dialogue: 0,0:13:07.46,0:13:10.30,Default,,0000,0000,0000,,"what is learning?" Fundamentally, \Nphilosophically.
Dialogue: 0,0:13:10.54,0:13:13.87,Default,,0000,0000,0000,,"What is learning?" We see that we\Nare learning, children are learning,
Dialogue: 0,0:13:13.87,0:13:16.24,Default,,0000,0000,0000,,everybody is going to school, \Nwe all are learning.
Dialogue: 0,0:13:16.24,0:13:18.36,Default,,0000,0000,0000,,We think that machine learning\Nis learning,
Dialogue: 0,0:13:18.47,0:13:20.50,Default,,0000,0000,0000,,but what is learning really, right?
Dialogue: 0,0:13:20.64,0:13:21.99,Default,,0000,0000,0000,,"What is understanding?"
Dialogue: 0,0:13:22.44,0:13:26.54,Default,,0000,0000,0000,,What does that mean?\NWhat does the word "mean" mean?
Dialogue: 0,0:13:28.07,0:13:31.39,Default,,0000,0000,0000,,What is thinking? We keep saying
Dialogue: 0,0:13:31.39,0:13:32.88,Default,,0000,0000,0000,,"Oh -- I'm thinking about this"
Dialogue: 0,0:13:32.88,0:13:34.80,Default,,0000,0000,0000,,What are you doing when you're thinking?
Dialogue: 0,0:13:34.97,0:13:37.94,Default,,0000,0000,0000,,So, today I'm going to show you an\Nequation of thinking.
Dialogue: 0,0:13:37.98,0:13:40.84,Default,,0000,0000,0000,,Okay? So, it'll be fun -- I don't claim
Dialogue: 0,0:13:40.84,0:13:42.76,Default,,0000,0000,0000,,this is - THE - equation of thinking,
Dialogue: 0,0:13:42.76,0:13:45.07,Default,,0000,0000,0000,,but I'm trying to get to that plot point
Dialogue: 0,0:13:45.07,0:13:46.93,Default,,0000,0000,0000,,where we start thinking about thinking,
Dialogue: 0,0:13:47.86,0:13:49.17,Default,,0000,0000,0000,,and not just think.
Dialogue: 0,0:13:50.30,0:13:54.48,Default,,0000,0000,0000,,"What is creativity?" Now, creativity is,
Dialogue: 0,0:13:54.48,0:13:57.82,Default,,0000,0000,0000,,if you look at an artist, or a musician,\Nor even a scientist,
Dialogue: 0,0:13:57.82,0:13:59.36,Default,,0000,0000,0000,,we create new inventions
Dialogue: 0,0:13:59.36,0:14:01.42,Default,,0000,0000,0000,,out of the knowledge we have,
Dialogue: 0,0:14:01.42,0:14:06.33,Default,,0000,0000,0000,,and innovation is a manifestation of\Nthe knowledge in a certain form.
Dialogue: 0,0:14:06.33,0:14:09.40,Default,,0000,0000,0000,,Right? A poet creates, \Na musician creates --
Dialogue: 0,0:14:09.40,0:14:11.20,Default,,0000,0000,0000,,so what is creativity?
Dialogue: 0,0:14:11.20,0:14:14.78,Default,,0000,0000,0000,,And the last question I have, here is
Dialogue: 0,0:14:14.78,0:14:16.88,Default,,0000,0000,0000,,"What is consciousness?" Right?
Dialogue: 0,0:14:16.88,0:14:20.21,Default,,0000,0000,0000,,So, ultimately, if you look at movies like\N"I, Robot",
Dialogue: 0,0:14:20.21,0:14:23.73,Default,,0000,0000,0000,,the word "I" from the robot is\Nnot really about
Dialogue: 0,0:14:23.73,0:14:27.11,Default,,0000,0000,0000,,the robot's great abilities at\Nmundane tasks,
Dialogue: 0,0:14:27.11,0:14:29.09,Default,,0000,0000,0000,,but really it's about the "I" in it.
Dialogue: 0,0:14:29.67,0:14:34.41,Default,,0000,0000,0000,,"I am a conscious being", and now what are\Nthe consequences.
Dialogue: 0,0:14:34.48,0:14:36.23,Default,,0000,0000,0000,,Right? So what is consciousness, and
Dialogue: 0,0:14:36.25,0:14:39.19,Default,,0000,0000,0000,,can we have sentient machines at the\Nend of the day, right?
Dialogue: 0,0:14:39.52,0:14:41.86,Default,,0000,0000,0000,,So, we won't go there today,\Nmaybe we'll see
Dialogue: 0,0:14:41.86,0:14:43.72,Default,,0000,0000,0000,,if we have time we'll watch a video,
Dialogue: 0,0:14:44.25,0:14:46.95,Default,,0000,0000,0000,,but I'll try to cover the bottom three and
Dialogue: 0,0:14:46.95,0:14:50.38,Default,,0000,0000,0000,,see if we can find something interesting.
Dialogue: 0,0:14:50.38,0:14:53.83,Default,,0000,0000,0000,,So, learning. Learning is one of the\Nmost basic things,
Dialogue: 0,0:14:53.83,0:14:55.39,Default,,0000,0000,0000,,we all do learning all the time.
Dialogue: 0,0:14:55.69,0:14:58.22,Default,,0000,0000,0000,,-- at least we all claim to\Nbe learning all the time.
Dialogue: 0,0:14:58.48,0:15:01.54,Default,,0000,0000,0000,,So, really, I'm going to use language and
Dialogue: 0,0:15:01.54,0:15:03.66,Default,,0000,0000,0000,,not vision at first, but language as
Dialogue: 0,0:15:03.66,0:15:07.12,Default,,0000,0000,0000,,the basis for all the examples.
Dialogue: 0,0:15:07.73,0:15:11.05,Default,,0000,0000,0000,,So, learning really is many, many things:
Dialogue: 0,0:15:11.47,0:15:13.71,Default,,0000,0000,0000,,the first thing we learn, so, you know,
Dialogue: 0,0:15:13.71,0:15:17.45,Default,,0000,0000,0000,,the greatest example of a machine\Nlearning system, or an A.I. system
Dialogue: 0,0:15:17.66,0:15:18.78,Default,,0000,0000,0000,,is a human child.
Dialogue: 0,0:15:19.49,0:15:21.14,Default,,0000,0000,0000,,And all you have to do is just observe
Dialogue: 0,0:15:21.14,0:15:25.13,Default,,0000,0000,0000,,how a baby is growing up, how he's\Npicking language, how he's
Dialogue: 0,0:15:25.13,0:15:29.33,Default,,0000,0000,0000,,picking walking, how he's picking \Nswimming, how he's picking tantrums, right?
Dialogue: 0,0:15:29.73,0:15:34.19,Default,,0000,0000,0000,,And you learn so much about A.I. because\Nyou're looking at the real A.I.
Dialogue: 0,0:15:34.19,0:15:36.78,Default,,0000,0000,0000,,So what is learning? I want to use\Nthat example
Dialogue: 0,0:15:36.78,0:15:39.01,Default,,0000,0000,0000,,and see how we pick up language.
Dialogue: 0,0:15:39.64,0:15:43.15,Default,,0000,0000,0000,,If I use the word -- if I start\N-- imagine you're reading a novel,
Dialogue: 0,0:15:43.43,0:15:45.76,Default,,0000,0000,0000,,or imagine words are coming at you one\Nat a time:
Dialogue: 0,0:15:46.26,0:15:50.05,Default,,0000,0000,0000,,you see the word "united" - what do you\Nthink the next word would be?
Dialogue: 0,0:15:52.80,0:15:57.36,Default,,0000,0000,0000,,Right? "United States", \N"United Something" , whatever.
Dialogue: 0,0:15:57.36,0:16:00.64,Default,,0000,0000,0000,,then, [MIC CUTS], predicting. When we're\Nlearning,
Dialogue: 0,0:16:00.64,0:16:03.03,Default,,0000,0000,0000,,we are also simultaneously predicting.
Dialogue: 0,0:16:03.03,0:16:05.92,Default,,0000,0000,0000,,And this is one of the flaws in current\Nmachine learning:
Dialogue: 0,0:16:05.92,0:16:10.90,Default,,0000,0000,0000,,that we keep thinking that learning is\Nseparate, prediction is separate.
Dialogue: 0,0:16:11.44,0:16:13.58,Default,,0000,0000,0000,,We'll learn first, then we'll score.
Dialogue: 0,0:16:14.44,0:16:16.84,Default,,0000,0000,0000,,Right? But the human brain is not like that.
Dialogue: 0,0:16:16.84,0:16:20.38,Default,,0000,0000,0000,,We don't learn for sixty years and then suddenly\Nwe start behaving.
Dialogue: 0,0:16:20.38,0:16:23.88,Default,,0000,0000,0000,,We're constantly learning and \Nwe're constantly applying that learning,
Dialogue: 0,0:16:23.88,0:16:26.43,Default,,0000,0000,0000,,and that is one of the fundamental\Nreasons why,
Dialogue: 0,0:16:26.43,0:16:29.15,Default,,0000,0000,0000,,you know, I call the current model of\Nmachine learning
Dialogue: 0,0:16:29.15,0:16:33.27,Default,,0000,0000,0000,,like the [inaudible] which is never going\Nto become a data-flow architecture
Dialogue: 0,0:16:33.27,0:16:34.81,Default,,0000,0000,0000,,ever, right? So that is one of the problems.
Dialogue: 0,0:16:35.19,0:16:38.28,Default,,0000,0000,0000,,So imagine what we're doing now, we are\Npredicting what will come next.