Kai-Fu Lee: China, Silicon Valley, and the Dual Visions of AI | Town Hall Seattle

everyone thank you for coming this evening welcome to the collective Seattle and Town Hall inside out my name is Alex mano and I'm really just appreciative to all of you for helping one of my dreams come true which is a lot of wonderful people learning about coolship space so thank you and we are just so thrilled we have a social club downstairs on the first floor it's a place to learn connect play engage build relationships we do lots of events like this for people like you we also do rock-climbing serve amazing food and beverage got great Wi-Fi people get stuff done during the day host good meetings and I just want to say we are thrilled to have Town Hall hosting this evening and with that I'll hand it over to Edward well you have Mike are you thank you so much well thank you all so much for being where you name is Edward Wiltshire I'm the curator of lectures at Town Hall Seattle the collective is a phenomenal host for us while our building is undergoing renovation over on the first Hill yeah I have a feeling there's actually probably a lot of you haven't been to Town Hall programs in the room tonight yeah we're so honored to be hosting dr. Kai Foley for this conversation this is some of the most important stories in science in business and technology going on in the world and he's one of the best voices to be covering that so we're honored to be hosting him for those of you for whom this is the first town hall event we hope you'll stay in touch with us we are producing programs all around Seattle while our building undergoes renovation and we'll be opening back up early next year the best way to see what we're doing at a glance is to pick up one of our monthly calendars here's our October calendar with the beautiful vish object sing the seek Captain America on the cover he'll be doing a program next week and we have a whole bunch of programs more on technology and science as well as politics and the arts and music so keep an eye on everything we're doing but tonight I would also like to thank our partners in promoting and building out the concept for this event the world affairs council so to say a little bit about their work and then to introduce our speaker formerly I'd like to welcome Harish Abed who is on their board of trustees thank you so much [Applause] good evening everyone once again welcome for this evening's event I was given strict instructions that I have just one minute to go through with the introduction so and I have a this tendency of drifting away so I'm actually going to read out from a sheet which I hope takes one minute so we are very pleased to be partnering with Town Hall it's been one of our very interesting partners hosting events with them and tonight's discussion is with with world with dr. Kai Foley who's a world-renowned leader in an expert in artificial intelligence so my name again is Harish wet I am on the board of the World Affairs Council as a quick introduction to the world affairs council for people who do not know you know our mission is to increase global IQ and awareness in Seattle we have a lot of programming around that space just to give you an idea the kind of programs we have community programs I'm going to read out a few upcoming events so you'll get an idea of what we do so on October 18 we are hosting a breakfast discussion with Microsoft president Brad Smith on global privacy and cybersecurity trends on October 24th we will welcome Pulitzer Prize winning New York Times reporter David Sanger for for a look at the rapidly evolving world of cyber conflicts and on November 13th we have Seattle mayor Jenny Durkin who will talk with us about the challenges and opportunities for Seattle as a global city you can find all this information and more about all our programs and beyond on our website WWE world affairs dot o-r-g we also do a lot of other programs one of them includes the K to 12 teachers and students to ensure that our classrooms our global classrooms and ready we bring the world Seattle by working with the Department of State so we have a lot of global exchange programs and we are lucky to have a strong citizen diplomat diplomats in greater Seattle that help us welcome about 800 international delegates to Seattle every year so we are very happy and proud of that so we welcome you to stop by the information desk Nathan and Laurie around will be happy to share information answer any questions you might have but now to the main event of the evening you know it is my privilege and honor to introduce tonight's speaker dr. Kai Foley who truly doesn't need any introduction so he's a prominent figure in the Chinese internet sector a founding director of Microsoft Research Asia then corporate vice president at Microsoft and later he was the president of Google China early in his career he was faculty member at Carnegie Mellon where he got his PhD in computer science and then moved to Apple computers presently he is the chairman and CEO of sinawe ssin ventures so as we know artificial intelligence is undoubtedly one of the most dynamic and revolutionary topics today he discusses this in his new book AI superpowers China Silicon Valley and New World Order's I found it profoundly intriguing and highly recommended you know if you as you walked out you know you saw the selling these books it's worth reading I think we are very fortunate to have dr. Li with us this evening to share his insights on how artificial intelligence is already impacting our lives and disrupting the world at a massive scale and how developed and developing countries are dealing with this dramatic disruption dr. Lee please welcome thank you well thank you for coming and an evening it's been a while since I've been to Seattle so you probably don't remember me so I want to introduce someone you know a lot better to tell you about artificial intelligence and why listen to this so this was synthesized by an artificially intelligent President Trump by an AI program let me be specific the end by an AI program that attempted to simulate President Trump you can type in any text it will speak in his voice in Chinese or English so in I wanted to give you the example to show you how far artificial intelligence has come and the fundamental technology that you see in the little gray area is how it works so for those of you not familiar with AI I'll just very briefly explain it it is basically a brain like architecture that has billions of numbers potentially and it is trained on huge amounts of data and it learns correlation among those data to do things like prediction decision classification and synthesis at a superhuman accuracy so this is why it's going to disrupt our entire industry and society but for those of you who are worried about science fiction and what if it's become super intelligent and cyborgs take over the world not to worry it has a very very serious limitation this capability this neural network deep learning can only do one thing at a time that is if it does speech synthesis for President Trump it does that very very well if it does stock trading it may do that very well it determines what Amazon ads to serve to you it will do that very well what face - to recognize faces of all of you who came in the room it will do that very well but it cannot cross it doesn't have the human ability to understand cross domains strategically plan create or even doesn't have common sense so in that sense it's not so in one sense it's like a prodigy that does one thing super well but it doesn't really know how to do anything else so we don't have any fear that this will take over the world unless there are ten more breakthroughs and when that happens I'll write another book okay but but for now I know there are books out there that say that and the reason I wrote this book is to tell you those books are wrong there are also other books written by historians and philosophers who paint a very negative picture and I want to tell you that's not the right way to go and those are the reasons I wrote this book the story of this book was I felt that as an artificial intelligence researcher businessperson and investor I have a unique 360 degree view on what's really important and it's going to change the human future of humanity bring about opportunities we never thought possible and challenges we never thought possible so I talked to some book publishers about writing this book and they told me that historians write better books like that and philosophies philosophers write better books like that and but they said chi foo if you will put China in the title this will sell so that is why you have the title that you have but fortunately my publisher and I found a very good wave to wave China into this book and into this talk as you will hear in the next 45 minutes so let me first tell you a little more about how AI will change the industries so if AI is single domain at a time and AI changes ái takes a lot of data then obviously the first industry that will use AI is the internet because we accumulate so much data every day we we are lab rats for Facebook and Amazon every time you click that's something that contributes to Facebook's intelligence about what you as a person and people like you like every time you click on Amazon to buy something or not to buy something that contributes to Amazon's knowledge and they eye about what you and people like you might buy or might not buy therefore allowing them to serve you ads on things you're most likely to buy so internet AI has created all these internet companies and they've all become in IAI companies so today's biggest AI companies are basically the seven Giants they are Google Facebook Amazon Microsoft and China's Baidu Alibaba $0.10 I would say these seven companies probably have the world's 90% of the best AI engineers okay researchers they're still many are in the universities but the best engineers are probably captured by these seven companies because they offer high pay stock options and a lot of data if you ever meet at AI scientist just say oh have so much data that I want you to look at and they'll say I want to work on it and and who has more data than Internet companies so of course they are fortunate to have so much data to lure the scientists the scientists build more technologies and products for them they make more money and then they get more users more scientists so that's the loop of these seven companies becoming very very powerful is something that's both a blessing and a challenge to the future so other than internet who else has data well every business has data banks insurance companies have data and this afternoon I was in Starbucks and they actually have a lot of data your cars every time you scan that's a signature of what you bought where and when so those companies used to treat data as a cost center right Bank feels like it has to archive all the customers day that just because for legal and accountability and those kinds of reasons so every year millions of dollars goes into the data center to store that data but today that data suddenly becomes goldmine because for bank by accumulating all those data from an individual customer it is then able to able target products at the customer if you have a new CD or a new mutual fund you know who are the customers who might buy it and how to sell it to them because you know what they've done in the past and if you want to check on credit card fraud whether someone will repay a loan or asset allocation those are all things you can do much better having that data as a goldmine and you can improve for example one of the companies we fund called fourth paradigm what they did was they went into a bank and said we want to implement AI for you and the bank was scared said well I can't put that in any of my mission-critical things and they said well how about you know your bank spams me all the time with text messages trying to sell me Bank products they said yeah he's as well I can target your spam better and they said well how will you do that this is on your customer profile and list profile I can estimate what customers are more likely to convert when you send that message and the bank said okay there's no downside after all it's just spam mail and I'll call a spy I'm sure the bank has a better term for it but we think it's spam right then the bank tried it and then 65% increased conversion so before they send 10,000 messages maybe a hundred people buy it now 165 people buy it suddenly the bank made a lot of money on just doing something very simple without taking any risk then they increasingly bought more of their products so you can extrapolate that to other domains of businesses so wave one and wave two are both using big data ai and what about something more than big data or wave three is actually digitizing the physical world the data that used to be transient in the world for example watching this lecture it's gone after my talk can now be captured and turned into training data to make something very smart for example those two gentlemen back there are capturing my video so they probably can use my video to build a new video with me or do something that's different and it's actual examples of that includes things that you're in Seattle so you probably all know Amazon go right that's an example of digitizing the physical world they put sensors and cameras in the store so they recognize people who walked in what items they picked up what eivin's they bought and then they just walk out without having a cashier or any salesperson it becomes an autonomous store on the basis of things that were transient namely what happened in the store is not what used to be not stored now they not only store it but they turn it into understanding the customer and also targeting products to the customer as well as checking out products the person took with them so that's an example of digitizing the physical world another also Amazon example is Amazon echo the fact that it's capturing the speech of what you said and there's a lot of useful information that it can do not only giving you what you want delivering the music you want played buying the product you want bought but also understanding you better through capturing the physical world so it is essentially adding eyes and ears to the AI so it's now can now has eyes in years the fourth wave is when the AI actually has hands and legs feet and fingers so it can walk and manipulate and move that doesn't mean that robot actually has physical two-legged are two-legged it could be running on the base on wheels but whatever it takes to get a job done so where can that be applied well the biggest application is clearly autonomous vehicles that's the huge one that we see way mo and uber and Tesla and the bunch of companies we invested in I'm making great progress in and that will change the world in much more dramatic ways than earlier examples because when you think about autonomous vehicles they are not just a button on the Tesla you push and let it drive for you for a few minutes on the highway that is not the way autonomous vehicles will be the ultimate autonomous vehicle will be the shared economy electrical vehicle and autonomous vehicle all merged into one so that none of us have to buy cars anymore until you know that the car is the worst investment that you probably remain because you buy it 96% of the time it's sitting idle depreciating 4% of the time it's giving you some purpose taking you from place a to place B but if there were an uber that is available in 30 seconds much faster than your parking and always available and never misses and probably one half the price and and safer better driver than you would you still buy a car right and when that happens it changed everything changes people will tilt towards an autonomous environment where more and more cars autonomous when cars are autonomous they will start to talk to each other so that there will be much fewer accidents first a I will gather more data so they will learn oh that's a person that's a person so they'll get better and not accidentally hitting people and cars will talk to each other about where the location they are so they could miss each other by one centimeter that you're inside probably scared to death but it's actually really really safe and the next thing that will happen is humans will be disallowed to drive on public roads that's inevitable because just like you can't ride a horse on the highway that's endangering other people that is what the future will be for those of you who love to drive just like their farms in which you can ride horses there will be there will be speed ways where you can still drive a car that's probably 20 to 30 years away it's not that imminent but it is also definitely inevitable because I think as you know the critical mass of these autonomous vehicles they become better and better and then we just want to get out of the way and enjoy a better life so what that will bring about is we save money because we don't waste money buying cars our lives a lot of you where people will die and the emissions will be much better because no gasoline and also all the deliveries will change all the transport getting you know today you have to pay Amazon Prime in the future that price will go way down you can get groceries delivered to you if you get lobsters delivered to you you know today the lobsters might take three days because a human has to sleep while driving it from East Coast to west coast in the future autonomous vehicles don't sleep so all that will disrupt the drive the entire automotive industry and and that's inevitable because you see capital going into it you see talent going into it the smartest engineers you know are probably working on autonomous vehicles you ask any company that's doing it with it that's the case and if smart people are working on it larger money among the money is going into it and also the other companies have given up you don't see any traditional car companies still fighting autonomous driving there they admit and they accept and acknowledge that one day this will come so it is inevitable and that will create potentially an operating system because if you autonomous vehicle can drive it can do all kinds of things it can be a robot so that may be the next big operating system after PC and Android the next one might be autonomous vehicle and robot but it won't be that easy because it's gonna take a long time so coming back to robots that's another big area we believe robots will happen first in industrial environments then in commercial and finally in the home because robotics is actually very tricky and very expensive autonomous vehicles similarly are very expensive at first price comes down you see the price of the original Tesla's and the Tesla's now is a very good demonstration that with volume price comes down so the first demonstration of robotics will be manufacturing replacing assembly line workers it's going to be in area some such as doing repeated tasks on things like dishwashers fruit pickers those will be robotics you want the robotic dishwasher who wants to buy one you know how you know what how that works is not like a dishwasher a robotic dishwasher is you take all the leftover from the table and just dump it in this big machine and all comes out with the dishes stacked Bowl stacked fork stacked all cleaned and also high temperature sanitized and all the garbage are divided into recycle bins now who wants that everybody wants that so it's okay good news we invested in such a company called dish Kraft and it's the product will come out at three hundred thousand dollars apiece so anybody still want it but what this demonstrates is just that it will get used at a high end first and the price will come down with volume so who's gonna buy from the high end imagine a very top restaurant I don't know what's a good example my memory of Seattle how about like Daniels and Bellevue right they probably employ five dishwashers right human dishwashers and one such machine amortized over a year and a half is probably about the cost of those human dishwashers and now you don't have to worry about people not showing up at work dishes piling up or people injuring themselves with the jet height high pressure jet hot water so and then once they sell to the you know three hundred Daniels in America then the price comes down to a hundred thousand they'll sell a lot more and it keeps going down eventually maybe in ten years you'll get to buy your one thousand dollar home robot that washes or your dishes so all this is going to happen and each wave will probably create on the order of five to ten person of GDP to our economy way for maybe higher may wave to maybe maybe the same I don't know but on the order of five to ten percent and also along the way you probably know this to displace five to ten percent of our jobs so that's the benefit and the challenges that we will face so this is last I'm going to talk about AI next is the part that Oh what more is that - for AI to work the following five conditions have to hold you have to have a lot of data the good tagging so it's not just random data you have to tell it this is John's face or Mary's face or this is a defaulted loan or a non defaulted loan and then only works in one domain and lots of compute power and then some AI experts if you've got these five things you're all set and and now it comes to okay who's who's ahead u.s. or China this seems to be the only question the media wants to ask me so I'm just gonna get it out of the way and answer it and then tell you why it's irrelevant okay the answer is u.s. is by far ahead in research you list the top AR researchers they're all Americans or Canadians no exception there's zero Chinese or actually zero non-american on Canadian they're all Americans and Canadians in fact not only is America way ahead in research America has been ahead in every core technology Silicon Valley has been at the center of the world and the only leader as the world the rest of the world simply uses its products but China has really changed in the last 10 years and this is an important part of the book of how China managed to challenge this us hegemony this challenge began with this cycle China has a huge market which attracted a lot of money and then VC's going to the market and they get smarter and help entrepreneurs the entrepreneurs are tenacious in China and I'll give you examples later and then they build great companies which have great products and business models that get more users and with more users more money comes in so as China's internet over the past 10 years went from about 150 million people to 800 million people the opportunity really has blossomed we're at a stage now where the total China Internet company valuation is roughly equal to the total consumer internet company valuation in the u.s. about one to one and I think China will get bigger for reasons I'm going to tell you and this is all pre AI okay but this is relevant so hold on to the thought I'll explain how it relates to AI this is before AI has come about in the last ten years China has basically created and equally valued internet mobile internet space and that has led to actually not just companies that are valuable but you read in the newspaper is that Chinese government protects local entrepreneurs and they steal American IP and block American companies from entering so that's how the values created but that is the wrong characterization I'm not saying none of those accusations are true in specific cases I think those may have been true but actually what happened was the Chinese entrepreneurs initially copied the American business model because the China market was at the time too small it was only 20 years ago when US Internet penetration was a hundred 50 times that of China so what are the Chinese entrepreneurs to do but to reference the models that exist in America that's how China's Google China's Amazon China's eBay came about initially I don't think that's equates to infringement of IP that's just looking at a UI and then and then copying it it is what Facebook did to snapchat okay some of people may not like it but it's not illegal so that was how China got started but that's actually an important today because where China actually got to in the last ten years is it has come up with a completely separate valid and and interesting and extremely valuable way of entrepreneurship I don't have enough time to go into every detail tonight so I'll just give you one example so these are kind of the differences but I'll give you one example that's very striking because I recently talked to the CEO of Yelp and the company that's closest to Yelp is a company called mate mate and damping and actually damping was originally a copycat of yep for those of you who are familiar with the products we all love Yelp Yelp is a good product it is developed on the basis people will eat the way they eat and here is a wonderful review guide that will help us better match what we like to eat so it's a great product it's a company worth four billion dollars wonderful but the Chinese company mate and EMP is something completely different it may have begun with concepts of Yelp and concepts of Groupon but it evolved into a fifty five billion dollar company valuation so it's some you know more than ten times that of Yelp how can that be possible because may times CEO dared to change the way people eat it wasn't you either way you eat here's a layer of software it was what does it take to change the way people eat and the specific goal he went after is food delivery to the home meal delivery to the home and that sounds like a pretty boring ugly technical problem but actually he went into detail and found out that people would order a lot more takeout delivery to the home if it could be delivered in 30 minutes while the food is still hot including cooking time and costs not more than 70 cents so most American entrepreneurs oh wow that's impossible let's move on and work on another AI technology right but the Chinese entrepreneur said no I'm going to work really hard use my operational excellence to iteratively take some money from the VC do some experiments see if it works if it works to do more of it if it doesn't work throw the way essentially iteratively using the i you know i iterate the livvie testing ideas until he got to the goal if he doesn't get to the goal he'll just keep going so are there are huge risks in taking that approach because they currently do twenty five million deliveries per day so if you're 70 cents is not seventy cents but two dollars and seventy cents then that means you're losing two dollars per delivery or fifteen million dollars a day imagine how large no bleeding that is to a company so that's the level of risk and that the Chinese entrepreneurs willing to take and then how did he eventually solve the problem through a combination of AI technology and really just brute force operational excellence he ended up building a fleet of six hundred thousand delivery people who wrote electrical mopeds that was the only cheap enough vehicle that could get to the seventy cents per per delivery and organized a battery changing station for these electrical mopeds because they're such cheap mopeds their batteries keep running out so he had to build all that and manage the six hundred thousand people and these six hundred thousand people are not your usual assembly line worker or uber driver they're usually people who cannot get those jobs there because how that's the only way you can get to the seventy cents per delivery yet you have to make sure your customers are safe these people are are are innocuous and then they can be trained to have present your brand and company and be courteous to customers imagine the training and the management it takes for these extremely minimum wage paid people to be riding on the electrical mopeds with the changing batteries how much pain that is and the risk you're taking if you don't get to seventy cents because you'll bleed to death with the amount of money you burn every day and on top of that of course there's AI there's computing the algorithm of who should take which order to which home change batteries when so that's a complex equation you can imagine it's actually like uber but tougher because uber is getting a person from place a to place B here you're have to go get the food and delivery and maybe get more food along the way deliver more so it's a more complex algorithm there's also incentives involved it's like uber has surge pricing right you all hate that right this has surge pricing - what if there are not enough people to deliver the food right now well you incent some people so that they come you they get a little thing on their screen that says your hourly wage has been doubled will you please come get up and deliver the food so that's the complication right that's the chinese-style and why would a Chinese company go through take that much risk to build such a heavy ugly risky company the reason is when you have a lot of copycats around you when copying is not frowned upon you need to build a business model that cannot be copied so what's the business model that cannot be copied you know an idea like a Groupon and Yelp can obviously be copied but if you have six hundred thousand people and operational excellence and it would cost someone ten billion dollars to replicate and kill you then you have something that's almost uncopyable so they did that no one could beat them so they won got God listed publicly in Hong Kong fifty five billion dollars but I'm going this is the first audience I'm going to make this prediction the game is not over Alibaba is going to spend 10 billion dollars to rebuild that Network and we will see who wins at the end we I don't know but this is the competition in China and then and that leads to businesses with impregnable business models and that's something that I think Harvard Business School case study will study one day and revere even though it doesn't have the brilliance and the light bulb of a Steve Jobs it leads to models that are arguably more sustainable so hopefully that's clear how it's a different model worth studying and my book will give you a lot more details and many other examples so because of that spirit and that business model in the past eight years China has gone from copying from the US and then inspired by the US but and then leapfrog for example those of you who use WeChat probably know it's much better than what's happened agree those of you who use Weibull probably know it's better than Twitter ok maybe not in the diversity of content but in the product design and usability right and then there are a lot of others who our investment is probably better than Quora and the Taobao is better than eBay and so on and so forth but that was only the second stage now China has reached the new stage where are all these companies that I am NOT going to describe to you because it would just take too long because none of them are based on us inspirations anymore their brand the Chinese entrepreneurs have been trained in the first phase and excelled in a second phase and now they're using the Chinese business model to create brand new business models in the third phase and that's why China is a force to be reckoned with and respected there may have been wrongdoings in the early stages there may have been issues there may have been scrappiness there may have been the lack of experience in innovation but today it is every bit evil to the American companies and many of the Chinese companies are being copied from China to Southeast Asia and other and other places so that has led to entrepreneurs who are extremely hard-working tenacious and they know how to raise money because you know may Tran has raised billions of dollars telling stories and eventually delivering and also incredibly hard-working so 996 means I am to 9:00 p.m. six days a week and this was a slogan of a certain Chinese unicorn that said we value work-life balance come work for us we're only nine nine six that company's also now listed in Hong Kong worth tens of billions of the holders okay so coming back to AI so now we've got these great companies great entrepreneurs and they're ready for AI because AI is all about iterating and getting more data so they're the perfect group of people and companies to use and expand AI so the third thing that China has is more money is going into China in they're more money in 2017 invested in China ai than USA I and the equivalent speech recognition company is worth more before worth less and now worth more than the equivalent US company and if I could brag a little bit San ovation ventures we were the first to invest in AI only four years ago these companies were all founded between the last two and four years and they're currently five unicorns valued at over 21 billion and these are pure AI companies I'm not giving you internet companies that's gone AI these are core air pure AI companies that do nothing but they I so you can see I I don't know if we can find that much value in Silicon Valley today in terms of AI companies so that's why China has gone really far ahead with the entrepreneurs the engineers and the business models and the data which is very important also AI is going through a transition in the early phases maybe five to ten years ago AI is about early adopter about researcher driven about PhDs who are so rare without them there's no company so the expert is the most important during those phases advantage went to us for creating AI companies but in the past five years as technologies like deep learning matured we've entered the era of application where it's more about finding where AI can be used and how to build an AI product and sell it to the right people and collect a lot of data and build a valuable company worth money and there the advantage would go China and that is an important shift that's important to recognize so some of you might say hey wait a minute is research not important anymore well of course research comes in cycles the next breakthrough might shift the advantage back to the US but but if we look at the last 62 years of AI there has been one breakthrough and that's deep learning 10 years ago we have not seen another equivalent breakthrough in the last 10 years so to predict the cycle is coming in the next 2 3 4 5 years it's possible but maybe a little overly ambitious and we talked about data being important this is a typical graph on the right five to four different algorithms each performing differently but when you have more data they all perform well so it kind of goes to show you can have your super researcher but if another country or company has more data they will usually beat you and if in the age of AI data is the new oil then China's the new Saudi Arabia and in China has depth and China has breath breath means more users period depth means per user there are more there's more data data like food delivery like the matron example data like payment data like shared bicycle rides there are ten 53 hundred times more than us multiplied by a three times more user advantage so the mobile payment is an important one to talk about Chinese mobile payment exceeded the China GDP that would seem incredible what that means is people don't carry cash anymore in China people don't use much credit cards in China which means those of you who have not been to China please be careful when you go because you might be stranded because if you bring out this cash no one will take it a lot of guys a lot of people won't take it and the credit cards some the larger stores will take it but there are many places you can't use it so for the Chinese though this is a huge convenience what this means is 700 million people are connected to each other fully connected not to merchants but to people anyone can pay any as little as 15 cents I can pay someone and then the Commission is basically zero okay it might be one tenth of a percent at times but generally it's zero somewhere between zero and one tenth of a percent compared to credit card companies which are really the parasite in the American economy they are not really adding value this stuff works you guys should get rid of credit cards but it's going to be hard it's going to be hard but what this really means for the Chinese economy is the following things first with respect to AI this is a ton of training data people spending pattern is so much more valuable than they're clicking pattern that's obvious right clicking doesn't necessarily mean you're committed paying definitely does so getting everybody's payment is important that's why Alibaba and 10 cent I think the company still had room to grow but it's not just their data it's also data that's that if you have a chain of convenience stores you're also collecting data knowing more about your customers because they used mobile payment to pay you so you have that track record that record as well so it's great for data upgrade for AI it also will turn a savings economy to a spending economy it's so easy to pay it also gives brings huge efficiency because you don't have to take that credit card and swipe and do anything with it also it's very advanced I mean one of the things I hate when I come to the u.s. is when I use my credit card they'll think I'm because I suddenly appear in a different country there might be fraud they'll call me and ask for my pet's name and what high school I went to right in China actually you know how it's done when you use mobile pay also triggers come up so Alibaba intention was there oh this might be someone's who stolen another person's phone but what do they do they actually don't harass me with questions they say please scan your face and then I just scan my face and it said oh that looks like a food order goes through so you might say what if I hold up a picture and do that it doesn't work because they'll say open your mouth turn to the left say this so they've come up with ways it's so much more convenient than remembering you know what's my favorite sports or something like that so it's tremendous convenience that's afforded to China as well and a lot of data for AI and one last thing is great for entrepreneurism because earlier you have to accumulate a lot of eyeballs in order to build an internet product and then figure out monetization later remember that most company and that company started that way they just got a lot of users and said well well well here's our business model we'll figure our business model later but you don't have to if you have mobile payment if you have ten just 1,000 users you can charge them ten dollars each and you might break even for the first month in your operations because it's so easy to pay people are inclined to pay and and another time or in the book there are examples of how the Chinese paying patterns are changing dramatically so that is a big deal and that that is data advantage for China the last advantage is a government advantage here's where China is often misunderstood again a lot of media and some politicians would portray Chinese government's as giving a ton of money to a Chinese company so it would compare advantageously to an American company again I'm not saying there are not examples of that but I'm telling you in AI that is not the case all the companies the five unicorns we funded are all privately funded they didn't take any government money in fact the government didn't fully jump into AI until about a year ago so all that came in the private sector but when the Chinese company the government does come in I have to acknowledge they add a lot of value I'll give you three examples of value they add the first is just setting the tone because China has a strong government when the central government says AI is important people listen so banks are more open to buying AI local governments are more open to doing things related to AI and then that there are environments that that creates one of the companies we we funded found that after this plan came out more banks willing to buy their products so that is one positive thing setting a tone this plan actually doesn't come with any money this plan is just setting the tone okay a lot of it is misunderstood in the West the second thing is that China has generally attacked new your toilet Aryan policy what does that mean it means when the new technology comes out the Chinese government will say let's not regulate it let's see what happens and then if problems happen they'll regulate it if big problems happen they'll shut it down I'll give you three examples mobile payment in the u.s. if someone were to say we're gonna do mobile payments we're going to ignore Visa MasterCard American Express we're gonna compete against them well you can imagine what's gonna happen next the credit card companies are going to lobby they're going to claim you know technology companies can't be trusted only credit card companies have experience right you can't trust Facebook with your money you can trust Visa you can imagine all those messages will go up to Capitol Hill and then there will be different hearings and you know who knows if that will get approved or not in China the government basically says well $0.10 Alibaba doing things like credit card let's give them a chance and then when they prove trustworthy let's give them more chances and three years later the credit card companies are gone so that's what happened but it doesn't mean Chinese government doesn't regulate there are cases where China regulates like p2p lending that is setting up a platform where any of you can loan money to any of the other of you and that obviously there are lots of potential risks and they try to roll it out and then let people do it and found there's a lot of fraud so then they regulated it by requiring license there are also examples like cyber currency which is very hot here a little bit controversial in China when they let it go out for a while but pretty soon they found it became really the love of the fraudulent activities fake icos even in the countryside you know poor old village ladies were buying icos and i said well how are we gonna deal with that they just shut it down so right now cyber as currency is not legal ICO is not legal in China so it's not unregulated it's just techno your Tillet Aryan give technology a chance regulated only when necessary that is an interesting policy that's worthy of studying don't have to adopt it here but interesting to study and then finally the last thing is infrastructure building so private companies fund private AI companies privately seized fund private AI companies however the government should do things that private companies cannot do so what might that be building how about building a new city the size of Chicago that has autonomous driving roads built-in or building a new highway that has sensors to make autonomous driving safer or a new part of a city ten square kilometers with two layers one layer autonomous driving one layer normal driving or another city where is also downtown two layers one layer top layers humans and pets and bicycles bottom layer is cars and that means the cars won't hit people and it also means the lighting will be very standard so the Tesla accidents like that won't happen so which way is right we don't know see the Chinese local governments are like little entrepreneurs they're each doing a little startup and they try new things very expensive billion-dollar projects but the one once something's proven to work it'll be standardized and used nationally so I think that's a very powerful weapon now some of you may still say hey that's still disadvantaging Chinese entrepreneurs because you're helping them get technology out there faster but one could also say President Eisenhower when he built the u.s. highway had the same intention and same effect so anyway I'm not a policy expert just explaining what actually happens in those cases just as a side note the US trucker's Union appealed to President Trump and secretary Chao not to let autonomous trucks be tested on highways so imagine what the impact of that might be is that us which is ahead in autonomous driving is kind of pulling back giving less chance to the one part of autonomous I mean that can be launched the fastest because highway is a lot easier than local roads in downtown a logical government ought to let trucks and highways go first but one that's concerned with truckers jobs may not but one has to weigh the value of truckers job versus the technology edge I mean if I were if I were the decision-maker I would let keep the technology edge and then finding ways for truckers to become retrained to other jobs or even give social welfare subsidies but but it seems in this case one is pushing forward once pulling backwards so to end this section if I were forced to give a zero-sum score I would say in implementation and value extraction China is a little bit behind today after all it's only been 2 years but in five years I think China generally will be ahead the only exception is in business AI because China business data is not well data warehouse compared to u.s. banks and hospitals so that's the score but coming back to the my real argument is that there is no zero-sum game here why is there a zero-sum game the Chinese money is funding the Chinese companies selling to Chinese consumers the American money is funding American companies selling to American consumers they're not even in the same market there is not even a consumer that would consider Chinese AI versus u.s. AI there are parallel universes so there is not a zero-sum game what the America government wants is American people to have a better life make more money feel happier the Chinese government wants that for the Chinese citizens they're not in a zero-sum game this is not a land grab fighting over our Santa Loren between you know Germany and France or fighting over oil in the Middle East this is not AI is not a limited resource where the Morea you're a US has the less China has this is not a zero-sum game both can learn and grow they're not even you know after the same markets and customers so this so but everyone asks this so I have to do it so the forces now I think there are many so that I think the key point now is that with US and China as dual engine driving a eye forward AI will make a lot more progress the faster progress the mobile on internet because you've there are two engines not one engine driving it the seven giants are hiring training a lot of people with a lot of data a lot of the funds are out there Softbank has a hundred billion dollars and new technologies are emerging with more apps that are coming so what this means that is that AI is arriving probably faster than PwC estimates which is fifteen point five point seven trillion dollars added to the global GDP by the year 2030 I'm usually a more optimistic on predictor so I give you here a more pessimistic number but even this number is huge sixteen trillion dollars that is the GDP of us I'm sorry of China plus India so that's how much in just a mere twelve years this will bring about but AI will also cost a lot of unprecedented challenges you in the US have seen so much of privacy security and bias I won't go into these areas they're covered a little bit in my book but I think the biggest issue is increased wealth inequality this graph shows pre pre AI days 1990 to 2015 the top two percent of the u.s. wealth earners and the bottom ninety percent which is the majority of the people you can see the two percent exceeded the bottom ninety percent in the year in 1995 and the gap continues to grow one could could attribute to that possibly the ICT computer internet mobile revolution that created a lot of Technol tycoons and also reduce the average income of many middle class and lower middle class workers and AI will only exacerbate that because imagine if monopolies or or powerful companies used to hold their monopoly or a powerful position by what by technology by brand by resource by entry barrier or by customer affinity but now when you have AI you actually have AI to reinforce the strengths of your product because your product is trained down more data so how can someone compete against you as a new entrant when they have less data so that means that the wealthy will get wealthier what about the poorer or that comes to my main other point in my book is that job displacement represents the largest threats that will that that will happen and if we think about what the single area single domain prodigy AI can do that AI can one domain do a brilliant job better than people that means the repetitive tasks and people will be replaced by AI as well the routine jobs as will the optimizing jobs in something like 5 10 15 years some jobs are reasonably safe but but but there are not that many of those jobs and these things are not future projections these are headlines from Wall Street Journal Financial Times there are robotic hamburger flippers that are being developed replacing one by one the individual worker this is a machine made by one of the companies we fund it is a pastry scanner so you know you go to pastry shops a tray full of tasty croissants and then a cashier hits the price of each one now computer vision can do that and then this the price comes up it's charged against your phone this is already running in Beijing we were so happy with the result we not only invested in the company that built the machine we invest in the pastry shop because they showed us the P&L and you can see the cost of a pastry shop the pastries cost nothing to make that and they're made in the central kitchens the cost of running a pastry shop is the real estate plus the people and the people are not needed for pastry shops they don't I mean they don't provide much service they're just there to hit buttons and the machines can do it much more quickly and the cost of this device you know as you know Amazon go is what ten million dollars or something to create this is eight hundred dollars a piece and it's a one-time cost and and you don't need a cashier ever so that's the speed at which this AI tidal wave is coming and they're not just coming at one machine replacing one cashier one burger flipper level they're coming faster this is another company we invested in it is called f5 future store and you see there's no people in it when you walk in it scans your face and then you have a wall of icons the icons are all food that you can buy beef noodle a drink a salad and then the robot makes it where's the robot the robot is here see the store is this wide and then this part is the robot it's not actually not a robot this is a giant machine that's got the broiler fryer and everything and it's got packages of things and mixes them together and then an average meal is about two dollars and compared to McDonald's which is five dollars so think how many Chinese users will prefer genuine Chinese food at 40% the cost of of a McDonald's from Kentucky Fried Chicken so what happens to job displacement in that case well have we displaced jobs of course we have to the extent that these types of stores take 50% market share in a Chinese fast food while McDonald's and KFC will surely have to lay off half of their workforce right but yet it's not as though McDonald's or KFC laid off their workers it's through an industry disruption and that's the power of the eye here's another white-collar example city has issued a warning to that it would replace with the actual use AI to replace about 10,000 that there are 20,000 operational staffs that is essentially do quantitative work because machines are better at it that's one on one displacement what about this as an example Totti how is a Chinese company it's a seventy five billion dollar Chinese company media company you've probably not heard of a seventy five billion media company for a long time this is a company that has no editors it is well it's a little scary but I have to tell it like it is you is Facebook newsfeed on steroids it is super optimized to maximize your number of minutes on this news app I am not endorsing this app but I am saying that it is it has an average per user time of 72 minutes per user per day so imagine the total number of minutes that Chinese users read news if 50% are reading this type of news or American users reading Facebook newsfeed that is so much less traditional media or traditional news site that they read therefore traditional new sites and media will have to lay off half of their editors I mean we all know a lot of great editors no self-respecting traditional media would ever hire a robot editor we know that however they're not hiring them but they're being squeezed from the automation ai side now of course I personally believe news feed and Towton should do a much better job than they do then optimizing minutes per user and that will evolve but I think I'm using it just to illustrate that without any voluntary displacement of editors editors jobs will get squeezed so what this means is there will be very few jobs available for many people looking for jobs and that is the very difficult situation however a little bit of relief is that if we think about what AI cannot do creativity is one of them there is actually another and that is compassion and when I use the word compassion to actually cover compassion empathy human to human touch and Trust and we just don't trust robots to do certain things and those are jobs that can be created in large quantities so if we now rethink about this picture if we have creativity as the x-axis there should be a y-axis that actually has the compassion empathy and human connection type of jobs and you move these jobs around and you'll see that while the lower left is in danger of being displaced there are actually a lot of job on the upper left that might have greater opportunities of hiring people that includes you know think about your own situation would you want a robot to be your beauty consultant wedding planner right how about tour guide and elderly companion concierge bartender social workers psychologists psychiatrists clearly not and what about jobs like nurses and the teachers are clearly not there are also jobs that will evolve the future of the doctor may be as follows the AI will do the analytical diagnosis and prescription should I take that as a someone's complaining that their jobs are all right I will I will finish this before the battery runs out the doctor's job will change as follows the diagnosis is probably better than by an AI tool but if doctors job remains as a human interface to tease out of the patient how are you doing how are you feeling are you feeling any pain did you have night sweats what's your family history how about your parents did they have this that kind of things it takes human connection to connect to get out of the patient and then provide the comfort confidence and that will increase the patient's chance of recuperation and survival we all know the that people when they have confidence they're more likely to be cured so that is the symbiosis of doctors and the analytical engine and also when we when we have that kind of a doctor is actually more like a nurse practitioner with an extra EQ training and the doctor's job suddenly becomes much lower cost to fill instead of ten years twelve years of experience maybe four years and then the cost of medical care goes down the number of doctors goes up they don't make as much money as they do now but there will be many more of them so that might be a path in the future for the for the bottom left quadrant to go to the upper left quadrant you can imagine someone in our customers support job going back for a nursing degree or a or a basically a consultative type of non nurse practitioner degree that is conceivable whereas going to become a scientist may be quite hard another type of job that might change is a teacher teachers a lot of a teacher's job may may change the components that are repetitive grading homework drills exams can be done by robot by the teacher can be more the one-on-one mentor to provide the help and maybe rather than a fifty to one student to teacher ratio will have five to one then teacher can then then then education can be brought to more people can be more personal more helpful and and that can be many more of them and well and if we want to go one step further why can't a parent who chooses to stay home as the homeschool teacher get paid as a home school teacher so that again because if you think about the one-on-one mentor who better to do that than the parent but the parent has a job they can't do that but if they're laid off in this job but if they said I'm gonna be a home school teacher I'm gonna prove I'll do a good job maybe there can be some subsidy take as a final example the elderly caretaker we know in America there are something like a million job openings as a elderly caretaker they are not filled why not there's not paid enough so what do we need to do or we need to tax the ultra-rich who get the top 2% and find a way rather than as ubi give it to everybody maybe to incentivize people to take the elderly caretaker job because as our longevity increases people over 80 need five times as much care as other people and there will be many more people over 80 in all countries and therefore that is a growing job category and more people would do it if only there would be higher pay and greater social status so those are the changes we need to think that job that is it's not just a simple matter of giving money to people like ubi it is about people having needing a job to feel fulfilled in their lives it is also about change of thinking about our social contract that our pay is not simply based on our economic contribution but also social contribution those are some big changes that need to happen over time but to conclude on this slide the four quadrants basically equate to lower left is going to be taken over by AI that is just not going to continue the lower right these are the scientists the columnist the script writers the the movie makers they will use AI as a tool to enhance human creativity so it's a symbiosis and then on the upper left AI will do the analysis human will provide the warmth the comforting and that's also a human AI symbiosis and of course upper-right is a safe job because it magnifies human creativity and compassion so when people tell you about a are job displacement is merely symbiosis that's only half right they tell you it's merely replacement that's only hard part of the picture so there is the blueprint of how human and will coexist in the book there's more details but this is the simplified version that I can describe here and so AI I think is just really like electricity it will empower a lot of things AI people are actually very open and sharing despite all this talks about trade war and things like that I know that Chinese and American uar researchers are really good friends they share everything there is the archive.org where all the papers are published instantaneously people even open share their their source code on github and in China we also have open data like AI Challenger and not and decamp so these are all efforts that shows the AI people are sharing in general and that we really don't want to be thought of as a weapon and certainly the one that we thought of as a zero-sum game so the global community I think is the same way how do we solve some of the problems about jobs and this and security I think there's a lot of wisdom globally and globalism can bring all this back if we kind of become more generous and share Israel might have some ideas about AI security Korea might have some idea about gifted and talented education Japan and Switzerland might have some idea about craftsman as a future job Canada and the Netherlands might have ideas about volunteerism so all this ought to come together to solve the crisis that we might face in the next 20 years so we've talked a lot about 20 years I'm going to conclude by thinking more like 50 years the 20 years will be very exciting very hard very challenging lots of opportunities lots of wealth and lots of problems to overcome but if we get through this 20 years the next 50 will be really amazing because if we really look back a lot of the social contract and the work ethics might just be wrong so in 50 years when we look back I hope that will let realize two things the first is that AI is actually serendipity it's coming here to take away the routine jobs so that we can think about what it means to be human because we're surely not placed on this earth by our maker to do repetitive jobs it's a little bit insensitive to think that in the next 20 years because we got to get over it but in 50 we will finally realize imagine if you were our maker you'd be very frustrated after thousands of years we're still stuck here thinking our job our work our life is about working we haven't gained any wisdom and Industrial Revolution has further forced us to think about jobs being the meaning of our lives so the maker is probably very very frustrated and said ok I'm gonna send you a I write a is gonna take away those routine jobs so you'll start to think what it is that makes us human and then lastly I think people worry about AI about all the challenges but I would say not to worry because AI is just a tool that we as humans possess the only ability to have free will and determine what the future holds for us so there will be an ending to the story of AI and it is we the humans who are right it thank you [Applause] dr. Kaku I have a question regarding earlier I mentioned that a cross Club we have seven superpowers big companies cross US and China I'm secure as owners from your perspective why so super color power didn't emerge from other great economies like UK Germany or Japan okay I think the markets are too small and I think the entrepreneurial ecosystem is very critical I think UK Japan Israel France Canada actually all have a lot of brilliant researchers but I think the markets are too small so it couldn't do the China model and then their VC and entrepreneurial ecosystem somehow didn't get going sometime it's just serendipity right in Silicon Valley there was the Fairchild people who left their child there are people who started kleiner perkins in Sequoia and they met each other and then the semiconductor people thought the that the PC people who taught the software people who taught the Internet people with all the mobile people with all the AI people so now Silicon Valley is vibrant and China I described how China got to be so sometimes you just have these it's like human civilization right somehow it formed in a few regions it didn't in other places and I think US and China were just blessed at the right times they were able to build the ecosystems and to rebuild it at this time by another country is too hard especially when they don't have the market sighs thank you okay thanks that was an interesting talk my understanding is that deep learning is pretty brittle I can think of two problems as adversarial examples and also that if it hasn't been trained on something then it hasn't seen something that it can't reason with the current methodologies can't can't handle that and your some of the things you were expressing were ideas that didn't seem to involve some sort of analysis so could you maybe elaborate on how we can get to that point or where you see the next step beyond deep learning or extensive deep learning can actually do what we would consider to be more intelligence than pattern matching sure sure thank you are you a recent AI researcher yeah great me too yeah no you're very much right I think deep learning it is very brittle this is why I said the more data the better this is assuming we don't come our brittleness then more data is bliss that's kind of an assumption of the China advantage and and actually in the future things might change some breaks through in transfer learning might take place so that models can be trained on very little data then China would lose that advantage that might be an area of a breakthrough in terms of the analysis people have come up with very clever variations of deep learning there is the can network for example as you know for that adversarial training there's reinforcement learning so when I speak as deep learning I actually mean the conglomeration of similar technologies that altogether can be mixed and matched so I did some simplification but those things as you know evolved after deep learning sort of to patch the holes so this is a projection of the next 15 years in the whole patching kind of way to glue together solutions for the jobs and opportunities that we have it's what it doesn't do is I don't think I have any idea self-consciousness can be done emotion common sense cross-domain reasoning very deep planning strategy creativity so I'm assuming these seven things don't get done but other things somehow get patched to while we're in that yeah you touched a little bit on political systems the difference between the China and the US yeah in developing AI but what about the fears of this on control yeah for instance Big Brother yeah knowing every single thing you do right and so I would like to know with these different political systems we have right what what are your solutions on having individualism having ability to do creativity yeah without constant control yeah I I think we've seen in a lot of cases including the Snowden case that often governments can resist the temptation to mix security with surveillance right it's really kind of the two sides of the same coin because more surveillance gives more security but it takes away freedom and I I think regardless of a form of the government whether it's a democracy or not the government does care about what about its mandate to rule and about what people's concerns are and I'd like to think whichever country there is a feedback loop that countries will get the feedback that some acts will be stepping over the line and that it will choose to self-correct I mean maybe that's being too optimistic but I understand where you're coming from and I think those are those are valid valid concerns and and I think I think government forms will evolve as AI evolves and it's hard to predict which is going to influence which I would say at this time I don't see China or other countries having the kind of data and control that a Big Brother truly has but I understand the concerns but I like to believe that anyone who's governing realizes that he or she is earning the mandate by understanding people's concerns and using in the feedback it may not be through voting it might be through some other mechanism one can have a glass half empty or half full so first of all thank you very much dr. Lee for giving this talk to us um yeah so I guess while I was listening to your presentation one of the things I noted was that a bunch of the Chinese companies you mentioned other than they were some of the major big ones like Alibaba and ten cent whatnot we're all a bunch of companies I frankly never heard of and so and I was thinking about that a little bit and so you know thinking through your talk it seemed like know maybe product market fit where some of these companies maybe they can find great product market fit within China but have much difficult more difficult time expanding beyond the Chinese market for whatever reason so that's sort of the cause eyes out there and so well you know I understand the likely the researchers involved will be having open communication and things like that from but from a business perspective do you think that it's possible that at least from a product point of view a lot of these companies will remain kind of bifurcated between like the Chinese set of the set of like companies that have Chinese product market fit and if I have companies I have Western product market fit yes that's a very interesting question you must be a product manager no misako engineer yeah okay so there's a new idea that's not yet in the book so you've triggered me to share the idea China is actually not one demographic it's actually 3 demographics the first we just think about the type of people and what they want the first layer is the obviously the earliest adopter the most educated wealthiest layer that's probably somewhat similar to the US obviously there are differences what you say about private market fit is valid but they're similar in types of if you just measure education and time spent on you know news versus games versus other things if you measure those attributes more similar to us there's a second group interestingly probably similar to but maybe 3 to 5 years ahead of Southeast Asia due to China's faster development and then there's actually another third that most of you never think about that's the poor part of China that's probably three to five years of ahead of Africa so here's the interesting opportunity think about the export of the these three groups abroad the first group is going to have a very hard time because developed economies already standardized on American technologies so every day they try they have a hard time so they don't try anymore the second grouping actually is already going to Southeast Asia and getting some success and then the third grouping is starting to go to Africa and getting some success and now also some work actually there's a little bit at first and second grouping that's going to the Middle East and interestingly that matches the low priority areas for American companies where I mean think about any American company that's high tech they're going to go after the highest our pool right the highest our police you and the easiest effort and the largest earning so that's obviously going to be us english-speaking countries developed countries Europe Japan and so on Australia and so on so us we'll take that and those are very big strongholds of us but China is now going to the areas where there might be a chance so and in fact in the third in the three-tier picture where the third tier is the innovative Chinese ideas where it's not based on American innovation or copycat many of those are the second and third tier so I would project a future world where China actually makes significant progress in Southeast Asia Islamic world and Africa cool I not thought about that way that's very interesting thank you it's a couple of sneha that's how we call China Thank You teacher careful I'm bill as the front of a I camp we are AI power to a compass that up or provide to the AI trainings for the developers so we do this my us for one year and recently we and in China said an office Paigey of course we're going to face in some of the competitions many of them like the bench for you invested like the show I change the poker so many of the companies and in China is a fair or maybe not a very success so I'm just wondering you know do you have any kind of like the strategies or suggestions that go to China marketer these are American companies yes by American Chinese but you said many companies didn't succeed some of them like a you know not very successful like you know eBay Amazon or you know Microsoft in China they're not very successful when of the reasons probably I think you probably before you know parallel universe yeah the local companies local teams probably don't have our authority to make decisions so we are thinking you know to let her look at him to run we provide as a tech technology the support yeah I'm not as you know looking for consultation suggestions yeah sure well as I said US and China have developed into parallel universes it is very difficult for someone to travel to another parallel universe and accept and expect to get acclimated very easily assimilated very easily and that is why you know people ask about can Google succeed in China I think the question isn't about government regulations or Google's determination or whether they should do censorship or not I mean those of course academic questions you can ask the fundamental issue is a Google Amazon or Facebook even if they could fully enter China at this point in time it would be too late because the alternate universe is already been built unless Google Facebook Amazon has a brand new product that matches the alternate universe such as way mo for example hypothetically so so I think for your company your companies may be similar you've been in the u.s. too long you're not deeply rooted enough we have actually looked at and funded many of this sort of cross-border companies they don't generally succeed because the cost of communication is too high the cultural differences is too large and who's in charge is too complex because the cross-border companies with founders like you who have u.s. experience will tend to want to be the CEOs but love living in America and don't want to move back and then yeah actually as you said to succeed in China if China is your market you actually need your CEO decision-maker and the largest shareholder in your founders to be based in China I think that is the issue I I think I think the I think the greatest value will come from that doesn't mean the parallel universes don't have independent values I think someone for example take myself as an example I was a parallel universe traveler in 1998 with a lot of knowledge from the US when I went to China I found that those knowledge was very useful to China but only if I don't live in the US and live in China and fully integrating the environment can I use that knowledge right I think similarly if I were to move back from China now knowing all that I know about the startups in China and the ecosystem I bet I can found or help found a lot of very cool companies in the u.s. now knowing what I know but I think is a commitment that if you are a parallel universe traveler that has learned a lot in universe a and you want to leverage that in universe B you have to a commit to living universe B and then take all that you have and learn all you can quickly about universe B and build a universe B company but leveraging the knowledge you have from University I hope kind of thanks by the way we woke you with a challenger okay great thank you so our artificial intelligence systems are rebelling and we are running quite late so we have only time for just this one last question with apologies to those in line thank you for being here so in your explanation of the job displacement and the way in which AI will become a symbiotic part of life I think about the notion of democratization of AI right making the powers of AI available to others so I work for a company called sensing and we're working on democratizing entity resolution for the average person because being able to effectively do that particular job well has been very expensive and hard to enter on so when we think about that there are a few core players that we talked about what do you see in terms of more players coming about that allow the powers of AI for whatever your purpose is for whatever your objective is to be made readily available to smaller companies who need to accomplish a job and want to flourish as competitors but may not have the the staff the gumption the money the data to do the thing right so you're talking about you know Google platform Facebook platform Microsoft platform etc right I think you know again in if you're if you're a glass-half-full you would say you're gasps have empty you would say these giants are going to dominate us if we use their platform we're stuck helping them make money and they're gonna dominate us we'll never be able to have our own IP but I think if you're a glass-half-full I would say that today there is quite a bit of competition among those three plus Amazon which added AI to and I think we if we are a little bit optimistic and hopeful as long as there's competition they're going to be forced to be fairly open with us I'm one of I mean we're I'm one of you now right where the weak ones just think about how open tensorflow has become in light of the competition from Amazon and Microsoft and Facebook so I keep telling my friends from those two weaker three weaker Giants that they got to keep pushing so that Google will keep tensorflow open I mean if you look at you know Google is about as a benign a giant as you can get however if you look at Google's behavior in Android it was very very open but when it dominated the world it kind of started to pull things in so this is I think unavoidable business behavior when some company gets too powerful nearly monopolistic they will be predatory to even their developers Google as benign as I have seen but it's even unpreventable for them so I think the only check and balance is for other giants to work really hard to give them a run for their money and then we can hope for a multi-platform world in which we smaller developers can keep using what they have I do think in the long term it is impractical for small companies to to manage the kind of stack that they have managed so using it right now I actually think tensorflow is relatively open so I think it is okay to use any of those platforms but but there are a few tricks built-in tensorflow voting thus favored google ultimately so they do have a final switch they can turn at some point so I think we have to just cautiously be optimistic and hope multiple platforms exist and of course final thing I'll say is depending on your particular application there might be enough open source that you can cobble something together I do think there is another possible platform potentially emerging with all the developer community of the academics and weaker people like ourselves if we continue to contribute maybe something like Linux can emerge again as an open form of AI I'm currently a little bit skeptical of that but I think we have we should give it a try thank you very much

Channel: Town Hall Seattle

Views: 18,880

Rating: 4.828125 out of 5

Keywords:

Id: Cl17Oax7I-s

Length: 89min 26sec (5366 seconds)

Published: Fri Sep 28 2018