Duration 40:53
16+
Play
Video

Building the future of artificial intelligence for everyone

Greg Corrado
Software Engineer at Google
+ 2 speakers
  • Video
  • Table of contents
  • Video
2018 Google I/O
May 8, 2018, Mountain View, USA
2018 Google I/O
Video
Building the future of artificial intelligence for everyone
Available
In cart
Free
Free
Free
Free
Free
Free
Add to favorites
32.36 K
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speakers

Greg Corrado
Software Engineer at Google
Diane Greene
CEO at Google Cloud
Fei-Fei Li
Chief Scientist of AI and Machine Learning at Google

Greg works at the nexus of artificial intelligence, computational neuroscience, and scalable machine learning. He has published in fields ranging from behavioral economics, to particle physics, to deep learning. Greg has worked to put AI directly into the hands of users via products like RankBrain and SmartReply, and into the hands of developers via open source releases like TensorFlow and word2vec. He leads several research efforts in advanced applications of machine learning, ranging from natural human co

View the profile

Fei-Fei Li is a Professor in the Computer Science Department at Stanford, and the Director of the Stanford Artificial Intelligence Lab. In 2017, she also joined Google Cloud as Chief Scientist of AI and Machine Learning. Dr. Li’s main research areas are in machine learning, deep learning, computer vision and cognitive and computational neuroscience. She has published almost 200 scientific articles in top-tier journals and conferences, including Nature, PNAS, Journal of Neuroscience, New England Journal of M

View the profile

About the talk

In this Keynote Session, some of Google’s leading minds on artificial intelligence and machine learning discuss their vision for a future where artificial intelligence can improve the lives of everyone.

Share

Hello, who's interested in AI? Me to 33 Okay. So I'm I'm the moderator today. I'm Diane Greene and I'm running Google cloud and on the alphabet board and I'm going to briefly introduce are really amazing. Guess we have here. I also live on the Stanford campus. So I've known one of our guests a long time cuz she's a neighbor. So let me just introduce him. First is Faith a doctor FEI FEI Li and she is the chief scientist for Google Cloud. She also runs the AI

Lab at Stanford University division lab and then she also I founded Sailors which is now a ai4all which you'll hear of a little bit later. And is there anything you want to add to that? I'm your neighbor? That's the best and so then the other so now we have Greg Corrado. and I Actually, there's one amazing coincidence both fei-fei and Greg were undergraduate physics majors at Princeton together at the same time. And didn't really know each other that well when the A-Team Percy

none neither of us in computer science and then rejoin later only once we were here anyhow, so Greg is a principal scientist in the Google brain group. He co-founded it and more recently. He's been doing a lot of amazing work in health with neural networks in machine learning. He he has a PHD in Neuroscience from Stanford and so he came into a i in a very interesting way and maybe he'll talk about the similarities between the brain and what's going on in the day. I would you like

to add anything else sir. So I thought since both of them have been involved in the AI field for a while and it wasn't, you know, it's recently become a really big deal, but it'd be nice to get a little perspective on the history, you know yours Envision in yours and Neuroscience about any I and end and how it was so natural to Fred to evolve to where it is now and what you're doing and start of Science of human civilization. This is a field of all these 60 years of age and it started with a very very simple but fundamental Quest is chemistry

think and we all know thinkers. That's all leaders like Alan Turing challenged Humanity with that question can machines think so about 60 years ago. Very pioneering scientist computer scientist like Marvin Minsky John McCarthy started really this field. In fact to John McCarthy who founded Stanford say I live in the very word artificial intelligence. So where do we begin to build machines? That's like at looking Inward and ourselves and try to draw inspiration from who we are. So we started thinking about building machines that resemble human thinking and when you think about human intelligence

you start thinking about different aspects the ability to reason ability to see the ability to hear to speak to move around make decisions manipulate. So a I started from that very core from Deschanel dream 60 years ago started to Deliberate as a field of multiple subfield which includes robotics computer vision natural language processing speech recognition and then important development happen around the 80s and 90s, which is a sister field called machine learning started to Blossom and that's the field combining statistical learning

statistics statistics with computer sign and combining the quest of machine telligence, which is what a I was born out with the tools and the capabilities of machine learning AI as a field wind through an extremely route for productive blossoming. Of time fast forward to the second decade of the twenty-first century. The latest machine learning booming that we are observing is called steep learning. Which has a deep root in your size without let you talk about and so combining deep learning as a powerful statistical machine learning tool with the quest

of making machines for intelligent whether it's to see or is it to hear or to speak? We're seeing this Blossom and last I just want to say three critical factors converged around the last decade which is the two thousands in the beginning of 2018, which are the three Computing factors. One is the defense of Hardware that enabled more powerful and capable Computing. Can lose the emergence of big date of powerful Theta that can drive the statistical learning algorithm and I was lucky to be

Lost myself in some of the effort and then the third one is the advances of machine learning and deep-learning out with it. So it's convergence of three major factors brought us boom that we received today. And Google has being that's taking all three areas. Honestly earlier than the curve most of the f word started to Eva in early 2000 and as a company, we're doing a lot of work from research to products. And it's been really interesting to watch the Divergence and exploration in various academic fields. And then

the reconvergence as we see the ideas that are aligned in space is it wasn't so long ago that feels like cognitive science Neuroscience artificial intelligence even things that we don't talk about much more like cybernetics work really all aligned in a single discipline and then they moved apart from each other and explore these ideas independently for a couple of decades and then with the Renaissance in artificial neural networks and deep-learning. We're starting to see some reconvergence some of these

ideas that were popular only in a small community for a couple of decades are now coming back into the mainstream of what artificial intelligence is what Pattern recognition is is really been delightful to see but it's not just one idea. It's actually multiple ideas that you see that were maintained for a long time in fields by cognitive science that are coming back into the fold. So another example Beyond deep learning is actually reinforcement learning. So for the longest time, if you looked at a university and you were

looking for any mention of reinforcement learning whatsoever, you were going to find it in a in a psychology department or cognitive science department. But today is we all know we look at reinforcement learning as a new opportunity as a something that we actually look at for the future of AI that might be something that's important to get machines to really learn it completely Dynamic environment in environments where they have to explore entirely you stimulate so I've been really excited to see How this convergence has happened back in the direction from those

ideas into mainstream computer science. And I think that there's some hope for exchange back in the other direction. So neuroscientists and cognitive Sciences today are starting to ask whether we can take the kind of computer vision models that they say helped Pioneer and use those as hypotheses for how it is that neural systems actually confused power own biological brains see and I think that's a really it's really exciting to see this kind of exchange between disciplines that have been separated

for a little while. You know one little piece of History, I think that's also interesting is what you did Faye Faye with imagenet which is a nice way of split explaining, you know building these neural networks where you labeled all these images and then people could refine their algorithms fight go ahead and explain that just real quickly about 10 years ago that the whole community of computer vision, which is the subfield of a I was working on a Holy Grail Quest problem of object recognition, which is you open your eyes. You can

see the world full of objects like flowers chairs people, you know, and that's the beauty block visual intelligence and intelligence in general and to crack that problem. We were building as a field of different machine learning models. We're making small progress, but we're hitting a lot of walls and the one nice I started working this problem and starts thinking deeply about what is missing in a way. We're approaching this problem. We recognize this important interplay between data and statistical machine learning

models. They really reinforce each other in very deep mathematical ways that we're not going to talk about the details here lies. Ation was also inspired by human Vision. If you look at how children learn it's a lot of learning through Big Data experiences and exploration. So combining that we decided to put together a pretty epic the effort of we wanted to label all the images we can get on the internet and of course we Google search the lot and we downloaded billions of images that used

crowdsourcing technology to label all the images organized. Into a dataset of 15 million images of organized in 20 mm categories of objects and put that together and that's the image that project and we democratize the to the research of world and release the open-source and then we starting 2010 we held an international challenge for the whole day. I Community called image that challenge and one of the teams from Toronto, which is now at Google won the image that challenge

12 and a lot of people think the combination of and the people are named Aldo in 2012 was the concept of what Exactly. So yeah, and it's afraid you've been doing a lot of rain inspired research very interesting research in the end. I know you've been doing a lot of very impactful Research In The Hills area. Could you tell us a little bit about that? Sure. So, I mean, I think that's a flavor for how we can try to approach a problem the kind of machine learning and AI that is most practical today is ones where machines learn through imitation. It's an

imitation game where if you have examples of a task being performed correctly, the machine can learn to imitate that's in this is called supervisor and soap what happened in the image recognition case. Is that buy by faith, they building an object recognition dataset, we could all focus on that problem in a real Crete retractable way in order to compare different methods and it turned out that methods like deep learning in artificial neural networks were able to do something really interesting in that space that previous machine learning in

artificial artificial intelligence to go directly from the data to the predictions and break the problem up into many smaller stuff without having to being told exactly how to do that. So that's what we were doing before that. We were trying to engineer features or cues things that we could see in the stimuli that then we would do a little bit of statistical learning on to figure out how to combine these signals but with artificial neural networks and deep learning early learning to do those

things all together and this applies not only to computer vision, but it applies to most things that you can imagine a machine image. And so the kinds of things that we've done like with with Google smart reply and now smart compose we're taking that same approach that if you have a lot of Text data, which it turns out the internet is full of what you can actually do is you can look at the sequence of words so far in a conversation or in a in an email exchange and try to guess what comes

next. Right inspired machine learning and so forth. And so, you know, this artificial intelligence is kind of Bring into question. What are we humans? And then there's this thing up there called artificial general AGI artificial general intelligence. What do you think's going on here? Are we getting to the AGI? I really don't think so variety of opinions in the community, but my feeling is that okay, we finally gotten artificial neural networks to be able to recognize photos of cats, right? That's really great. Now can I

know but that's not enough kind of thing that's working. Well right now is this response were were able to recognize something kind of reflexively and we now have I believe machines can do pattern recognition every bit as well as humans can and that's why they can recognize objects in photos. recognition and that's why they can win it a game like go but that is only one small sliver tiny sliver of what goes into something like intelligence memory and planning and strategy and contingencies even emotional intelligence either things that are we

haven't even scratched the surface and so to me, I feel like it's really a lie too far to imagine it having finally crack pack pattern recognition after some some Decades of trying that we are therefore on the verge of cracking all of these other problems that go into what constitutes general intelligence. Oh, I believe yes or no Humanity has a tendency to unthumb to to overestimate of short-term progress and underestimate long-term process of eventually,

we will be achieving things that we cannot dream of but Diane and Greg, I want to just give a single example to defy AGI. So the definition of a g i a n is a introspective definition of what humans and human intelligence can do. I have a 2 year old daughter who doesn't like now. The I I thought I'm smart enough to scheme to put her in a very complicated sleeping bag. That doesn't get herself out of the crib just a couple of months ago. I was on the monitor

watching this kid 2 year old or for the first time she I was training her for not paying for it by herself. She was very angry. So she looked around figured out a weak spot on the crib where she might be able to climb out figured out how to unzip her complicated sleeping bag. I thought I scheme to do really in a tutu to prevent that I figured out a way to climb out of a crib. That's way taller than who she is and managed to escape safely and without breaking her. Okay. How about a GI equivalent

to my cat equivalent to my to a mouse if you're Shifting the definition Sure, but I think there are things that the cat is capable of an organism like from a behavioral level like the house has behave and how they respond to their environment. I think that you can imagine a world where you have something like a a toy that you know is for entertainment purposes that approximates a cat in a bunch of ways in that the sorts of behaviors that the human observe you're like. Oh it

walks around. It doesn't bump into things that meows at me every once in awhile. I do believe that we can build a system like that. But what you can't do is you can't take that robot and then, you know dump it in the forest and have it figure out what it needs to do in order to get to survive and make things work. Free videos capacity to help us solve. All our big problems is going to outweigh any kind of negative. Then we're pretty excited about that. I guess

like liking Cloud you're kind of doing some cool things with automl and so forth the believe of building benevolent technology for human use, right our technology reflect our values. So I personally and I know Greg's whole team is working on bringing a high to the two people into the fields that really needed to make a positive positive difference. So iCloud were very lucky to be working with customers and partners from all kinds of vertical industries from health care where we

collaborate to agriculture to sustainability YouTube. Entertainment to retail to Commerce to finance where our customers bring some of the toughest problem then they're paying points and we can work with them handing him to solve some of that. So for example, I recently we rolled out of my mouth as that is the recognition off the pain of entering machine learning. It's still a highly technical field. The bars are still high not not enough people are trained experts in the world of machine learning but yet

he already has so many to so much need who, you know tag pictures understand imagery. He's still says exemplary Envision. So how do we all serve that cause of need? So we worked hard and thought about this the sweet of her product called automl where the customer wheel Or the entry barrier by relieving them from coding machine learning custom models themselves. All they have to do is to give us the kind of provide the kind of data and concept they need here's an example of a Ramen company in Tokyo that has many

shops of Romans as they want to build an app that recognize the Romans from different Ramen store and they give us the pictures of robins and the concept of their store one store to store 3 at what we do is to use a technique of machine learning techniques. That's Google and many others have developed learning to learn as an amputee customized model for the customer that recognize Rodman's for their different stores. And then the customer can take them all. To do what they want. You know, I can

write a little C plus plus maybe some JavaScript. Could I absolutely we're working with teams that they don't have not even cplusplus experience. And then we have a drag and drop interface and you can use automl that way cuz I really believe that you know, there are so many problems that can be solved using this technique that it's it's critical that we share as much as possible about how these things work. I don't believe that these Technologies should live in Walled Gardens, but instead we should

develop tools that can be used by everyone in the community. And that's part of why we have a very aggressive open-sourced and to our software packages particularly in Nai and that includes things like tensorflow that are available completely freely and it includes the kinds of services that are available on cloud to do the kind of computer. Storage and model Sunni and serving do you need to use these things in practice? And I think it's amazing that we the same tools that my applied machine learning team uses to Ted tackle problems that were interested in those same

tools are accessible to all of you as well to try to solve the same problems in the same way and I've been really excited with how how much it's how great the uptake is and how we're seeing expanding to other languages of mentioning JavaScript quick plug for tensorflow. Jas is actually really awesome and you should probably run it on a TPU toe. So you're doing your building I mean with machine learning we're bringing into Market in so many ways because we do we have the the tools to build your own models the

tensorflow. We have the automl the brings that Tammy programmer and then what's going on with all the apis and then and how is that going to affect every industry and what do you see going on their clothes already has a suite of apis for a lot of our industry partners and customers from translate to speech to Vision to butcher based on models that we built a fox is a major partner with Google Cloud. Where are the recognized a tremendous need for organizing on customers imagery Theta to help customers. So they actually use

Google's Vision API to know that that's easily delivered to our customers through the robber as service. Play out in the health industry. I know you've been talking about their health care is one of the problems that a bunch of people are working on a Google in a lot of people are working on outside as well. Because I think there's a huge opportunity to use these Technologies to expand the availability in the accuracy of healthcare and part of that is because there's a

doctor's today or basically the weather in information hurricane in order to provide care. And so there's there I think there are thousands of individual opportunities to make doctor's work more fluid to build tools to solve problems. They wants all and to do things that help that help patients improve patient care. How many doctors are so unhappy because they have so much drudgery to do is this is this a big breakthrough? You know, when you go to the doctor, you're you're looking for medical attention, right? And right now huge amount of their attention is not actually

focused on the practice of medicine, but is focused on a whole bunch of other work that they have to do that doesn't require the kind of insights and Care in connection the real practice of medicine does and so I believe in machine learning and an AI is going to come in to help her through assistive technology do what they want to do better a system. No substitute for the speaking of human. Do you want to talk a little bit of you've been so you think this humanistic may I approach is so critical

to the first six? Years is a tie as more lesson Niche technical field. Where were still laying down scientific foundations, but starting this point on a eyes one of the biggest drivers of societal changes to come. So, how do we sync about AI in the next phase? What is the frame of mind that should be driving us has being on top of my mind and I think deeply about the need for human centered AI which in my opinion includes three elements to complete the human Center thinking the first of the month is really add event seeing AI to the next

stage and here we bring our Collective background from your science cognitive science, you know, whether we're getting to a GI tomorrow or or or in 50 years. There's a need for a i to be a lot more flexible nuanced. Faster and more unsupervised semi-supervised learning ways to be able to understand emotion to be able to communicate with humans. So that is the more human-centered way of advancing. Aai's site. The second part is the human Center AI technology in the application is that I love what you're saying that there's no substitute for human has this technology like all technology

is to enhance humans to augment humans not to replace humans won't play certain tasks will replace humans out of danger or our tasks that we can offer form. But the bottom line is we can use AI to help our doctors to help our disaster relief workers to help decision-makers. So there is a lot of technology in robotics design Processing that is centered around human centered and technology in application. The elements of human Center AI is really to combine the sinking of AI as a technology as well as the societal impact. We are so nice and I'm seeing the impact of

this technology but already likes I am sad that we are seeing the impacting different ways ways that we might not even predict. So I think it's really important and it's a responsibility of everyone from Academia to Industry to government to bring social scientist philosophers Los collares policymakers epices and and historian at the table and to study more deeply about AI social and humanistic gamepad as that is the three elements of human centered AI That's that's

pretty wonderful. And the end. I think we are Google alphabet or working as hard as we can to do humanistic AI, you know what we need to be careful about out there with a i and Regulatory. What are some of the barriers to you know, I think every company in the world has a used for AI in many many ways. I'm insist exploding and all the verticals but there are some impediment to adoption for example and financial the financial industry. They need to have something called explainable Ai, and could you just talk about some of

the different barriers you see to being able to take advantage of AI Healthcare? Yeah. So I think that there are there are bunch of really important things to consider. So one of the things is of course, we want to have machine Learning Systems design Anthony's of the folks that are using them in applying them and that can often include not just giving me the answer but telling me something about how that was derived the some kind of explainability still in the healthcare space. For example, we've been working on a bunch of things in medical

imaging and it's not acceptable to just tell the doctor that oh, you know, something looks fishy penis X-ray or just skin you have to tell them what do you think is wrong but more importantly you actually for that conclusion, why'd they can then look at it and decide whether they can cover or they disagree or a while there's a speck of dust there and that's what the machine is picking up on the good news is that these things actually are possible and I think there's kind of been this unfortunate mythologie that

Ai and deep-learning particular is is a black box. It really isn't we didn't study how it works because for a long time it really didn't work that well, but now that it's working. Well, there are a lot of tools and techniques that go into examining. How do you say systems work? And I think explainability is a big part of it in terms of making these things available for a bunch of applications. Add fire I think FiOS is a issue. We need to address in a I and I See Fire from Where I Stood two

major kind of biased when you do is just one is the pipeline of AI development starting from the bias of the data to the outcome of the bias. And we have here a lot of heard a lot about if this machine learning algorithm is fed with data that does not represent the problem domain in a fair way. We will introduce bias weather is missing a group of people state or or a firestick that to a skewed distribution of the things that would have consequences whether you're in the healthcare domain or Finance or legal decision-making. So I

think that is a huge issue. Very nicely that Google is already addressing that we have a whole team at Google were thing on fire true. I think it's important to the people who are developing. They isolationism bias and lack of diversity is also in some of our getting close to the end. But if you you know, where is AI going I mean how prevalent is it going to be? I mean, we look at our universities and these machine learning classes have 800 people 900 people, you know, there is such a demand every computer science.

Graduate wants to know, where is he going? I mean will every High School graduating senior be able to customize AI to their own purposes and and how will you know, how What does it look like 5-10 years from now? Susan from a technology point of view. I think that's because of the tremendous investment in resource posing a private sector as well as in the public sector now every many countries are waking up to investing I ask where are going to see a huge continue development of Technology you I'm mostly excited either at cloud or seeing what gretz team is doing

a I being delivered to the industries that really matter to people's lives and the work of quality and productivity quotes making it easier to use than educating them in and what's it going to look like? I don't you know. What do you predict? That's a really tough question because of the core of today's I still calculus and that's all going to change. So I think that from the kind of the tech industry perspective or from the computer science education perspective. I think that we're going to see Ai and ml become as

essential as networking. Oh, well, I'm going to write some software and it's going to be damned alone on a box and it's not going to have at ECPI connection. Right? Like we all know that you're gonna have a TBI connection at the end of the day somewhere and everyone understands the basics of the networking stack in the end. That's not just at the engineering of the level of Engineers. That's the level of designers of of of Executives of product developers and Leaders name thing I think is going to happen with machine learning and AI which is that designers are going to start to

understand. How can I make a completely revolutionary kind of product? In machine learning the same way that we fold in networking and internet Technologies into almost everything we built so I think we're going to see tremendous uptake and it becoming kind of a pervasive background part of the Technologies, but I think that in that process the ways that we use they are going live all so I think right now you're seeing a lot of things we're at some some spice some extra little coolness on a future and I think that what you're going to see

over the next decade is you're going to see more of a core integration into what it means to the product to actually work and I think that one of the great opportunities there is actually going to be the development of artificial emotional intelligence that allows products to actually have much more natural and much more fluid human interaction. We're beginning to see that in the assistant now with speech recognition speech synthesis, understanding dialogues and exchanges. I think that this is still in its in its infancy. We're going to get to a point where the product that we build

they interact with units in the way that humans find most useful to take out of the box and I spend a lot of time with high schoolers cuz I really believe in the future. You know, we always talk about AI changing the world as I always say the question is who is changing a I and II me bringing more human Mission sinking into technology development and thought leadership is really important not only important for the future of our technology in the valley with instilling our technology but also in bringing the diverse group of students and Future

Leaders into the development of a I so, you know, that's never a Google with all work a lot on this issue. And personally, I'm very involved with a I fall which is a non-profit. The edge of high-schoolers around the country from diverse background whether there are girls or or students of underrepresented minority groups, and we bring them unto AI campus University campus and the work with them on a I thinking and I studies and we're just completely committed to bringing all our best Technologies to everybody in the world and we're doing that through the

cloud and we're bringing these tools for bringing these apis and the training and the partnering and the processors and we're pretty excited to see what all you guys are going to do with it. Thank you very much.

Cackle comments for the website

Buy this talk

Access to the talk “Building the future of artificial intelligence for everyone”
Available
In cart
Free
Free
Free
Free
Free
Free

Access to all the recordings of the event

Get access to all videos “2018 Google I/O”
Available
In cart
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “Software development”?

You might be interested in videos from this event

September 28, 2018
Moscow
16
159
app store, apps, development, google play, mobile, soft

Similar talks

Daphne Luong
Director of Engineering at Google
+ 3 speakers
John Platt
Software Engineer at Google
+ 3 speakers
Rajen Sheth
Senior Director of Product Management at Google
+ 3 speakers
Fernanda Viégas
Research Scientist at Google
+ 3 speakers
Available
In cart
Free
Free
Free
Free
Free
Free
Zak Stone
Product Manager for Cloud TPUs at Google
Available
In cart
Free
Free
Free
Free
Free
Free
Sundar Pichai
Chief Executive Officer at Google
Available
In cart
Free
Free
Free
Free
Free
Free

Buy this video

Video

Access to the talk “Building the future of artificial intelligence for everyone”
Available
In cart
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
558 conferences
22059 speakers
8245 hours of content