Vivienne Ming | Machine Learning & Artificial Intelligence | SingularityU Brazil Summit

well hello I apologize right off the bat for the English but like any good American I'm barely literate in that much less Portuguese or anything else so bear with me and hopefully the language isn't gonna matter because what we're talking about is people the value of a human life and I'm gonna throw around some dollars and cents on occasion but that's not actually the value of a human life the value of a life is who that person could be and it turns out it's a question that comes up again and again in research and economics and neuroscience my field how can a person be and if we lived in a society where everyone had the chance to grow up to be the best person they could be what what would that world look like you have heard a lot of stories about amazing cutting edge technology I love the neuroscience talk because I actually do research on neural prosthetics myself but when you think about every one of these talks here today and all of the ones tomorrow think hard about whose life it's changing and how we bring everyone along for these amazing transformations James Heckman the Nobel prize-winner University of Chicago looked at how it pays back when you invest in a young child he specifically set up a school for little kids eight weeks old free education eight weeks to five years did this for single mothers in the state of South Carolina in the United States he paid exorbitant costs to create that school and run this experiment and what he found was investing in these kids pays off it pays off at 14% a year what other investment could you make it will give you a guaranteed payoff in the double digits every single year then in the life of a young child Raj Chetty the economist at Stanford published a whole line of research in this field one of my favorites recently looked at patents who patents new inventions who is out there inventing this new world for us all to benefit from and he found disturbing things that families from the wealthiest children from the wealthiest families had a vastly greater chance of creating a patent in their life not just then children from poor families but even those from middle-class families think about what that means how many innovations how many cures have been taken out of your life because we couldn't figure out how to give every child the potential to be the person they were born to be in my own work we actually looked at an array of experiments done over many many years that were able to do interventions on kids lives one of them in fact by the before-mentioned James Heckman and but a variety of them and we said what if we could do all these things that we know make a difference on every single kid 50 million children in the United States and what we found is that it would add an additional 1.5 trillion dollars a year to the US economy every year doing things we already know work but we don't do it similar interventions run in South Africa would add 60 percent to their GDP in India a hundred and ten percent so I took the time to run the model I didn't have the best data on Brazil but I did it here and our estimate is that it would come in somewhere between 70 and a hundred percent of GDP you have the opportunity to double the economic power of this country just doing the things we know work but it's not enough to just say that there's a problem what does it take to actually change a life and if you had access to incredible technologies that seem like they're only there for the elites what would it take to take those technologies my own personal specialty is artificial intelligence and actually make them available in a way which could reach everyone I'm going to share some stories about doing exactly that because I decided that what I wanted to do with my life was spend it making other people's lives better and we created a think tank to do exactly that so here are a couple of the projects we've worked on and maybe if we have time some new ones that we're running into in the future but let's start with this I'm going to talk about artificial intelligence AI can't solve anything it's not a magic wand and I'm not Hermione Granger I can't just spin out an AI and have it solve a problem it's just a tool an amazing tool but it's still just a tool every problem is a human problem and humans are messy and that means every solution I'm going to talk about every solution to every problem discussed by every speaker is going to be a messy human solution and if we're not willing to understand that truth we're not going to make a difference so let's dive into the idea of augmented intelligence I'm here to talk about education but I'm going to keep this pretty broad so instead of AI just being artificial what if the whole point of AI was to make us better to actively build a better person some of this you heard about in our neuroscience talk and the most extreme sense and it is true I am a professional mad scientist and I have every desire and intention to jam a whole bunch of really sophisticated equipment in your brain and make you smarter or possibly at least treat a variety of brain disorders but let's take it even more basic whatever invented something I had the pleasure in my very first company of developing a technology that could actually just listen to children talk to each other it could monitor students having discussions online about college courses and we actually found using this system at week one of a new course we could predict how they would perform in the class but the point wasn't to make the prediction the point was to make it not come true if they were going to fail the class what can you do to make a difference how could you intervene so the reason I throw this out is because as cool as that really sounded to me as a first-time entrepreneur this was before the rise of Coursera and Khan Academy and nobody wanted to fund the educational technology so we thought let's let's come up with another idea that would allow us to just prove how this system works so what would you do if you were at an amazing technology and you were a first-time entrepreneur so this is a look at there we go what we built and what you're seeing right now is what we might now call a deep neural network we built this before such things were in vogue and it was learning it was learning about faces so we turned it loose on Facebook and Flickr and a variety of other sources and it simply learned so none of these things that look like faces here are actual people in fact their words their words in a language of faces that this AI was learning on its own just by looking at faces on these websites and in fact you're only seeing one part of the dictionary there are thousands and thousands of words just in this section alone and then as you move higher higher level in the dictionary the words become about gender and race they become about courage and competence the faces of all sorts of different things so if you have this amazing technology that can understand faces even our abstract ideas of them what would you do with it well in this case we built an incredibly sleazy game called sexy pace and sexy face you go and our promise was you play the game we show you a random array of faces off of Facebook and you set the ones you think are sexy and our promise was for free we will find everyone on Facebook that you think is sexy and for $5.00 we'll find everyone who thinks you're sexy so as it turns out I'm not an incredibly awful person this is what's called a Trojan we make the promise and as soon as you play the game once the game confesses oh this isn't really the tawdry sleazy game it's a mind-reading game pick any face category and we can guess it so oh I don't know Southeast Asian muttonchops and the sense of home we well yeah have fun translating that well it turns out even making those really abstract judgments if you were really consistent in how you're making it our system could learn it I could learn what you thought of as competence it could learn ennui I could learn these ideas even if you couldn't articulate them now it turns out even that was a Trojan if you wanted to build a system that could return value to a human life I could think of no life more deserving of that than a young child who has lost their parents to warfare and as a band abandoned in a refugee camp and as it turns out there are a million of them quite literally about 10 years ago what we were working on this project I was approached by a group working for something called refugees United is a a program to United Nations to reunite orphan refugees with extended family members and they heard about what we were doing and asked if we could help and we built this system not to be a stupid mind-reading game a sleazy hook-up app we built it because when people made those judgments they trained our AI about how we perceive faces and then we built another version of it that pulled the faces out of the book that's what the UN had at the time a book full of a million photographs of little kids lost in camps around the world and instead of leafing through that book for days and hoping you didn't blink when your niece went by you could look on a on a tablet and you could select the kids that look a little bit more like your niece and in three to five minutes if she's anywhere in the world you found her that is the power of what AI can be it is an amazing tool for selling products for optimizing supply chains even for doing some more notable things like improving cancer diagnosis but it turns out tools like this can make a life-saving difference in kids in the most dire of circumstances and this is why I work in this space a lot of what you hear about AI can make it sound very scary and some of the things should be there are a lot of big implications of what AI is and how it'll affect our world but if you say no to it then you say no to this and you say no to this so this is scientifically proven to be the cutest kid in the world a scientific panel of two world-renowned scientists were put together and my wife and I agreed our son is the cutest kid in the world now we may sound biased but I'm a scientist so you know completely trustworthy six years ago my son threw up one Sunday and we didn't think anything of it and then it got worse and worse over the subsequent days and by the term he finally brought him into the doctor's office after everything been done on the phone with them for days they didn't even need to run a test you could smell it in his sweat it was sweet he had type 1 diabetes it's an autoimmune disease that destroys the cells that make insulin without insulin sugar can't get out of your bloodstream and inside your cells and you literally starve to death with a stomach full of food he had lost 15 pounds off of a 40 pound frame feel free to do the conversions yourself it was a lot and it was scary in a single week he couldn't stand up anymore and we rushed in and we spent four days at Oakland Children's Hospital up in the San Francisco Bay Area in the pediatric intensive care unit and it was brutal the one thing I can say is we got to leave as a whole family and not everyone did there were clearly other families they're worse off than us and that gives you perspective but it was still hard but it turns out the most shocking thing was a month later for that month my wife and I both scientists were collecting data every day data data data we recorded everything everything he ate to the gram every activity level did he have the sniffles that day was he grumpy we crashed Google Docs I hacked all of his equipment broke a whole bunch of laws it turns out and we were so proud of ourselves we had all of this data about my son and we we sent it to his doctor before we went in for his first visit they didn't seem to say anything so we thought oh we know what they want we printed out about an inch thick of spreadsheets of just ream after ream of data about my son and they were angry with us angry for wasting their time so just as a little aside if in the course of your job today you get a bunch of data and you think to yourself what the hell am I supposed to do with this your job is not long for this world think about how you're going to change your skill set so that you say thank you once again someone gives you a bunch of data but that isn't where medicine is today not in every day treatment and at that point six years ago that's not where endocrinology was and so they gave us a single sheet of paper and they wanted 15 numbers five days morning afternoon evening blood glucose reading an estimate of the sugar in the blood and that's it with 15,000 numbers and they wanted 15 hand scrawled they then eyeballed it for a few minutes and decided okay here's what we'll do for the next six months now I really like my son's doctor this has nothing to do with her or her competence but at the same time I sat there thinking you have got to be kidding me I build models of the brain are you telling me diabetes is more complex than the brain this is nuts so that day I started working on a system called jitterbug my wife and I met each other doing ballroom dance so all of my projects start off with ballroom dance names so we were doing jitterbug and and I hacked all this equipment I instead of sending it to the company's blackbox server then they were doing nothing with it I sent it to my own and I built a model and it turned out it could predict an hour into the future whether my son's blood glucose was going higher low now purely from a business perspective this is a multi-billion dollar market why was I the first person to try and build an AI for diabetes it was crazy how many families have gone before my son has challenges he was also recently diagnosed with autism and he has hard nights but we get a special reward that when he has a hard night and I spend it with him I get to tell him that because our family went through this because it happened to us and not someone else very possibly millions of people will be alive and that he was at the heart of that and that means something to him and it means something to me that we made a difference this is how I have dedicated my life and you know it isn't the case that a kid with diabetes or another chronic illness should be written off any more than those refugees or any more than the next story here so I was asked to join the board of a small company called the moja and they had a really cool technology they could take your mobile phone and without asking you any questions we could estimate your mood states your emotions overtime and they wanted help with the business and they wanted help with the technology and I said I would help you if you give me all of your data and let me do something with it and what was amazing is I said there's this disease called bipolar disorder it is also called manic depression famously characterized by elated creative highs and crushing depressions and I thought we could do something about this I had actually just read a paper as it chance would have it the night before about how you could take people with a genetic predisposition for bipolar and bring them into a lab every week and run a big battery of tests on them and actually predicts their first ever manic episode and I thought what if we could do this for everyone just using their phone and so the the next bit of serendipity is after I met with the founders of the company and I agree to join the board one of the employees at the timing competence came to me and said I have bipolar I overheard you talking you can use my data so much like with my son we had one single person at the beginning and we revised the models we updated everything we were working on and we built a system that could predict a manic episode by three to four weeks now I don't know how much any of you know about bipolar disorder but 25% of those severe sufferers will go on to kill themselves it leads to job loss and divorce and institutionalization these are people that have enormous potential to give the world but unfortunately even the standard drug treatments only work in about half of patients what if they knew ahead of time we could send a message to them to a trusted confidant hey high probability of a manic episode in the next week increase your drug dosing take a week off from work check in with your doctor and importantly we send a message to someone they love or sister a parent a spouse so they can be a part of this and they can help this is the opportunity to transform someone's life and we can do the same thing with things like depression or PTSD and take someone that has something to give but also has what we call a non neuro typical brain one that doesn't quite work the way the rest of our does but still has value and we can put those people back into society so they can continue to contribute and here's another one you know I learned how to do the face recognition work from my undergraduate lab my very first work in machine learning was this lab at UC San Diego where we were doing real-time lie detection off of video sponsored by the CIA so morally grey I don't know how you think of the CA here in Brazil but even in America you know they are taken as maybe a necessary evil so you know we spent the entire time we built this system thinking what else could we do with it and the funny turtle side story is that that lab spun off as a startup and then it had the clever idea of doing as you're seeing actually this is one of the cofounders of the company and you can see the faith the emotion recognition in real time as she's looking at the camera there they have this clever idea they did a real-time expression recognition during one of the u.s. Republican presidential debates back when I think many people were wondering whether half of them were even human beings and literally the next day the bidding war started between Apple and Facebook so if you have an iPhone 10 all the fancy new face recognition in there the the way that you can smile and talk and animate a cat on the other side that is 20 years worth of CIA sponsored academic research turned into a goddamn cat video which is what happens to all innovation in the end anyways they're all animates cats on phones but again I wanted to find something and I was proud of being able to do the work with the refugees that really meant something to me but I had another case that also has touched me and I actually mentioned it just a moment ago if I could take this and build it into a wearable pair of glasses in this case it was Google glass and they gave me an early release pair to experiment with and come up with ideas on what to do and so I could put this on and I guess right off the bat we could read all of your facial expressions see whether you're lying I could float your credit score's over your head or pull up your Facebook accounts or your Grindr accounts maybe whatever floats your boat I'm not interested in any of that it turns out that there's a group of people who can learn how to read facial expressions but they don't get it for free the way the rest of us do and these are autistic kids again amazing people competent what we call higher mid performing autistic children can grow up and become amazing contributing members of society they even have abilities that are exceptional for certain types of jobs but we don't treat them the same because they don't socialize the way the rest of us do so what if we could build a system for Google glass that could read your facial expressions right the emotion up on the screen so that they can learn how to read faces in everyday interactions in their lives and that's exactly what we did the gold standard today is still cartoon faces on flash cards but there's something that a cartoon face slacks which is purpose a mind a context the cartoon face isn't happy because it just got an award it isn't sad because you said something mean to it and so it turned out not only when we use the Google glass system that reads the facial expressions can these kids learn how to read facial expressions but it turns out it also teaches them empathy because in their lives before this didn't mean anything emotions happen and then they didn't happen and it was all random and now suddenly it made sense people were smiling because something nice happened to them people were sad because they've been hurt and they could learn what this meant what we call a theory of mind I'm enormous leap roud of the opportunity to work on these sorts of projects but I also recognize that sometimes they feel like specialized projects like we're focusing on people that have particular disabilities particular needs so let me instead focus on the broader story of Education and I think to understand how I view education you need to understand how I view how work is changing in this world there's someone who said recently that the world is moving faster than it ever has before and it will never be this slow again when you're thinking about what sort of people do we need as part of our community as part of the Brazil ten years now 20 years from now the goal here isn't to guess what that job will be or the specific skills we should be teaching people to do the purpose is to build people for uncertainty there will only be one job description going forward creative adaptive problem solver and in fact problem solver is even too directed via one job it'll be problem identifier I wanted someone that I can say you know I think there's an interesting problem somewhere way over there go find it figure it out and let me know if you need my help people like me will build a eyes that can look at a spreadsheet and do a risk analysis that can look at an x-ray and do a diagnosis that can read a contract and find the loopholes not only better than a lawyer but infinitely faster and none of the cases I just mentioned are coming they're here already the one about the contracts and the lawyers they read non-disclosure agreements humans versus a eyes the AIS were slightly more accurate but more meaningfully the humans took 90 minutes to reach each contract and the AI took 22 seconds I don't need any of you useless bags of organic material just to pull a lever for me no matter how cognitively complex that lever is I need you to explore I need a society of explorers we all need a society of explorers but the way we build them today is not that so calling human capital toxics asset as I am here in a short story I once wrote is not meant to be just mean-spirited it's meant to say if we keep raising our children such that we make certain that they score right on the right tests and they go to the right University and there all the right skills to do a job which will never change throughout their lifetime then we are headed towards a very bad future but it doesn't have to be that way here's one specific story now I'm hoping that none of you have Mis spent your life skateboarding but if you have shout out the name of this person okay so you're all much better behaved than many of my audiences so this is perhaps the greatest skateboarder of all time he won 59 world championships in a row from the late 70s to the early 90s and I actually had lunch with him I was a lovely lunch and I had a question not about skateboarding which I could care less about I wanted to know let's imagine it was your 40th World Championship what do you do immediately afterwards and he thought about it for a minute and he said well you know I probably go to the after-party and a drink a little champagne I don't know 20 30 minutes later I'm out back practicing new moves now in my short interaction with him I can tell you a couple of things one is he wasn't afraid of losing he wasn't practicing because he was afraid of getting beaten 40 times in a row kind of played the lesson out he also really likes to party so he's not I'm a social misfit I don't know why anyone ever goes to a party but he loves it he wasn't skipping it because he didn't want to be there he was practicing because that's who he is he's a fanatic this quality is something we discovered I was the chief scientist of a company we built a web database of a hundred and twenty-two million people and we found the crucial components that predicted the best performers and it didn't matter software developers salespeople designers it didn't matter and one of the things we found we called endogenous motivation that drive that comes from within exogenous motivation things like getting the bonus not getting fired praise from your parents getting the right grades it turns out that's a negative predictor of long-term performance if you're doing it for someone else you're doing it wrong and if you were raising your kids to do it for you or anyone else but them you're doing it wrong you are robbing them of their long-term potential turns out not only a crazy skateboarder but the best salespeople in the world you want to know who they are find out who makes the sale the day after the end of the sales cycle when nobody cares these are the most incentivized people in the world everything they earn is based on how hard they work and yet the best salespeople are the ones that are selling when nobody's looking they practice when nobody's looking they write code when nobody cares we need to build kids like that and in fact that's just one out of what we currently track 50 different constructs I don't even have them all on this slide but so instead of describing the individual ones let's just go to the basics general cognitive ability what we call metacognition labeled here is self regulated learning these are things like strategic thinking and self assessment a cognitive abilities like working memory span numeracy literacy social skills social skills turn out to be just as important for software developers as they are for salespeople they may be a little less common but when they have them it's just as important emotional intelligence mindset purpose believe it or not purpose is something we can measure and it is a huge predictor of positive outcomes and then all these constructs under creativity why are these things on this list 1 they are measurable we tend to focus our education system on things like do you know to factorize a polynomial or when was the war of the because it's easy to measure you got a right answer you got a wrong answer unfortunately it doesn't actually predict very much about people's life outcomes but these are measurable yes it helps to use the kind of tools that I'm about to talk about but they're all measurable they are predictive of long-term life outcomes and in fact sometimes you can demonstrate causally the role they play what do I mean by long-term life outcomes there's a paper in the Proceedings of the National Academy of Science last year just looking at five of these constructs in senior citizens in the UK as those five constructs increased the following things got better income wealth walking speed at age 65 social connectivity subjective well-being what psychologists call when they're talking about happiness and seven different biomarkers associated with health outcomes including genetic related ones like insulin sensitivity your genes the trigger to give you diabetes and because you're strong and qualities like this you don't get it that is amazing that's profound this is what we should be building in our schools this is what we should be hiring for in our companies and we should be committed to the lifetime learning of these meta learning constructs because they are what makes a better person now how do you actually do things like this well I've had the pleasure of being able to work on a number of projects in this space and here's an example of one this is Muse it's something that we built for teachers so I mentioned earlier how we worked on this project where we could record kids talking to each other and in one of versions of that we actually had microphones in a classroom with five-year-olds and they were chatting away and we thought what do we do with this so we're using all these fancy a eyes to analyse what they're talking about what could we really do that would make a difference who cares the most but they're these kids have an amazing life well their parents do their caregivers their grandparents their foster parents whomever that might be so we have this idea what if we just send a text message home every night and given what we learned we tell the parent hey do you have 20 minutes tonight here's how you spend those 20 minutes they were the biggest positive impact on your child's life and since then we've refined mused repeatedly it is one of these crazy mad science experiments it's out there in the world and in fact we're now giving away entirely for free either via SMS or an app and the whole point is we ask you questions about your kids we can analyze their artwork using very fancy a eyes and learn things about them and little notes that you make and from that Muse constructs an activity for you to do specifically with your child each night now what's most important about those activities and what we're doing with this system and others is I'm not telling you who your child is in fact I refused to tell you who can your child be that's what I'm interested in what will it take to get them there and why if you don't answer the why and you're working in the AI space you're cheating people you very possibly are lying to people because you really probably don't know what's going on inside your AI when you're talking about making recommendations about the life outcomes of kids you better be saying why so every night we send off a new activity for parents to do one of my favorite recently was I took our toilet apart with my daughter because it thought she needed to strengthen some ideas around problem-solving and exploration like a lot of little girls she gets a lot of messaging about you know not being in a meaningful effective agent in the world and Muse peaked up on that and bam we went hard at let's get dirty and it was great I loved it I didn't know how eyelet really worked how we took it apart and then I had to put it back together or get divorced so and my son actually for years now has been getting a lot of messages and activities for us to do around emotional support and emotional resilience I make no claims about Muse as a health tool but I am genuinely proud that had picked up on this trend long before his autism diagnosis ever showed up and we're working on ideas of this for young adults creating scavenger hunts for life experiences which will cause you to grow four versions of Muse to be used inside companies to actually help managers work with their direct reports so that you actually get them and you know what they need work itself should be a growth experience every job we assign someone should be partly yes about getting that job done and partly about who is the person that that experience will spit out on the other side in fact I take that so seriously that I say this technology has to challenge us we need to build a world designed to make us better we can no longer be passive about this and the way I define this when it comes to technology is not only that we're better when we're using it that's turns out I was about to say that's easy but then you look at a lot of technology out in the world I don't know it's pretty hard to argue that it all makes us better but there are some some that genuinely passed that test but that's not good enough we need to be better than where we started when we turn it off again and I know of almost no technology which passes that test very few but imaginal world designed to drive us towards our better selves where every experience is just for you and meant to create who you could be and it never ends it doesn't matter how old you are you always grow I'm over time but I'm gonna finish with this anyways I've founded a lot of companies I've had the chance to hire a lot of people and give them jobs I've invented technologies which have cured diseases or at least treated them and made people's lives better and none of it should have happened not a single one of those inventions or those companies or anything I've ever done should have happened I started with a life that was very privileged but by the time I was in secondary school I was already struggling and by the time I was in my 20s I was homeless and in 1995 I had a very long night living in my car with a gun trying to figure out why I should be alive I'd never done anything worthwhile and I had never really been happy despite the amazing best efforts of my parents and I came up with a reason live a life that makes other people's lives better that it isn't about me and it isn't about whether I'm happy it's about whether I'm serving my purpose and so what I'm asking all of you is as you go about your lives your companies the work you do philanthropic lis and beyond please be thinking about all of those kids kids and favelas outside Rio in villages outside go to car in my home in California kids that will grow up to invent the cure to the disease that might well kill one of your children and they will never have the chance to make that invention because of where they were born and because those fortunate among us didn't feel that it was worth investing in them what would the world be like if not a few hundred million people but billions of people all had the chance to contribute everyone this is the world I want to build it's the world I want Mike kids to grow up in I'm selfish our lives will be better and I got the chance to come up here and brag about some cool things I've gotten to work on but I did it because none of those things should have happened and my life slipped through the cracks and there were a billion lives or more out there slipping through the cracks today please leave here and go do something amazing give all those kids an amazing life thank you very much [Applause]

Channel: Singularity University

Views: 5,364

Rating: 4.6701031 out of 5

Keywords: Singularity University, Education, Science, leadership, technology, learning, designing thinking, future forecasting, Ray Kurzweil, Peter Diamandis, 3D printing, AI, artificial intelligence, AR, augmented reality, VR, virtual reality, automation, biotechnology, blockchain, computing, CRISPR, entrepreneurship, future, futurist, futurism, future of work, future of learning, genetics, health, healthtech, medtech, fintech, nanotechnology, robotics, talks, Machine learning, Prosperity

Id: 0nPUot7i9cY

Length: 43min 18sec (2598 seconds)

Published: Fri Jul 27 2018