HumAIn: Empowering Inclusion with Technology (Noelle Silver Russell)

Hi everybody, nice to meet you. Noelle Silver Russell. I just got married, so I'm getting used to that. That's why I had to do Noelle Silver Russell, like it reminds me to say the last name, otherwise I'm like, Noelle, how do you say it again? No, I'm just kidding, just kidding, honey. When he catches this on the replay. Super excited to be here. I'm going to actually do a couple things this morning. Let me switch over. Oh no! I might need someone to come and help me with a password. Thank you so much. If you could just take that long thing in there. Yeah, sorry. So, oh, here, we'll just put this back while you're doing that. So one of the reasons why I got, I don't know, I think they asked me, yes, they asked me to come to speak to you all. One was because of the wonderful way the world works. When you do something cool and you're excited about it, People, they like excited people. So one of my friends who saw me speak was like, "Oh my gosh, you should come do it over here." The same thing happens in music, in art. When you are good at something and then passionate about it and you aren't afraid to kind of take the stage, you get a chance to do it, well, as much as you want to. So today I travel the world Speaking about artificial intelligence, and I'll share with you as we get going. Yay! Oh, I think it's this one. Thank you so much. Oh, haha. Can you plug that in there? But one of the things that I want to share with you first, we mentioned Alexa in the beginning. How many of you know what Alexa is? I know, it's silly, right? This is just a practice, like, just get me in, you know, In the Motion. So good, most of you know Alexa, thank goodness. Well, I had one of those moments. I don't know if you know, um, or I've ever had this moment when you get an idea and you're like, this idea could change the world. Or you get invited to listen in on an idea and you're like, oh my gosh, this is the one, we could do this. And you get super excited about it. And then you have one of two choices. You could do that thing which often is scary and something you've never done before, and oftentimes something no one's ever done before. And so you could go and do it, or you could wish you had done it. And many times, I mean, I hear people all the time looking at TV, seeing some commercial, and go, "I thought about that 10 years ago." Our parents say it all the time, "That was my idea, I thought about that." Um, but they of course didn't do the building. Um, well, I was at Amazon, I was working as a cloud architect at the time, nowhere near artificial intelligence. This was just, I'd say, 7 years ago when Alexa— before Alexa was born. And I had an opportunity, I got an email, and the email was from Jeff Bezos. I know, it wasn't directed at me. It would have been cool, but it was like to the whole company. And he was like, we are looking for people who are passionate about this area to come and join this new team in artificial intelligence. At the time, it was not launched, right? Nobody knew it. As a matter of fact, at Amazon, we had just come off one of the largest failures Amazon has ever had, which is called the Fire Phone. I don't know if any of you remember it, but it was a cool device. But it was a cool device, but it failed. And everyone on that project was like, I don't really want to do another thing that's going to fail. I want to stick to things that work. And of course, at the beginning, this thing called Alexa did not work. Work, by the way. But I, I thought it would be really cool. So before I tell you why I thought it'd be cool, I want to introduce you to my people. This is why I do what I do. Um, these my gaggle of humans. I have 4 children. My firstborn son, though, has Down syndrome, if you're familiar with the condition. And when he was born, it was very interesting. I was a first-time mom, a young mom, never did any of the things you do when you think you might have a child with special needs. Like, I didn't know anything. Um, I have my child, it's normal birth. Only after do they like say, "I think something might be wrong." And in that moment, I know a lot— of course it's very nerve-wracking. Um, but it was interesting, the solution— they had my child in Knoxville, Tennessee, um, and the geneticist came out, old lady, like in my mind I remember her being like this, the though I don't think she was, but that's how I pictured her, like crotchety old lady. And she said, "You have two options. The world's not built for people like your son, but we've created little pockets for him to be successful. One is an institution, there's lots of them worldwide, and the other is you can give him up to an adoptive family who will take really good care of him." Now, I don't know if any of you know Latinas, but I'm a Latina, and I wasn't about to just be like, "Oh sure, I mean, take him. Good. I hope he has a good life." That was not the direction I chose to go. But I'm also a technologist, and so 17 years ago— he's about to be 18, which is crazy— but 17 years ago, I decided to embark on a journey of accessible technology, of how do I work for companies— not little companies, big companies— that would move the needle forward to make a world that's better for him. And that's him right there in the pool, Max. Um, so really cool. But so that was my lens, right? When I got this email from Jeff Bezos, my lens was like, oh my gosh, my son who's now 12, this could be cool. Now I am a cloud architect. I knew nothing about artificial intelligence. So I go to my manager, which you will do one day. You will get a cool idea and you'll go to your manager and you'll say, I want to do this thing. And they will look at you. Some of you will have awesome managers who will be like, "Rock on, go do your thing, be successful, go forth and prosper." But my manager patted me on the head, not physically, but I definitely felt it, like the, "Ah, you're so cute." No, you should, that's not for you. You're not an AI person. Oh, did I mention I didn't graduate from high school? Oh yeah, and I didn't graduate from college. I went to both though and put in very good effort. Just wasn't for me. So my manager was like, "No way! There's no way you're gonna be successful in a role like that." And I, being a naive middle-career person, was like, "Maybe, maybe you're right." A lot of us get advice like this from people we trust: our parents, our teachers, our— in this case, my first-line manager. But something in me said, "This is different. This is that one idea that if I don't do something, I'm going to regret it." And so I jumped. I jumped. I went around this person. I went to the hiring manager of the Alexa team. I said, "I know nothing. I'm completely transparent about my inequities as an AI professional." And he was like, "That's okay. We actually need people with the desire to learn because it's all new and an attitude of enthusiasm because the number one thing that makes AI successful is getting other people to use it. And that's what they needed, someone who was excited about it and could think about things in a different way. So I joined the team. That year, I built over 100 applications for Alexa. Today, some of my most popular applications, I mean, I have over 2 million unique users of the applications that I've built. I'm not a developer. I don't have a development team. It's just lowly old me on a Friday night coding. Some of you might be this person. It's— it was fun, but I built these applications, and at the time I built them, I literally was trying to think about, like, my son, for me, what would a mom with a kid with special needs want? Well, the year that Alexa was created, first I had my third child, then like the week Alexa went live. I might have had my phone with me, like, in delivery. Ew, don't do that. But it was Alexa, I was being born. I had 2 children that week. This is my dad, and he— the year we launched Alexa, something really interesting happened. Terrible, but trajectory-changing. He was hit by a car as a pedestrian. He's a Marine, so he's fine. Of course now, kind of. But he was running. He ran 11 to 15 miles every day. 68 years old, was running across the street. A distracted driver was, of course, driving and not paying attention. He was in a crosswalk and he got hit. And in a second, he aged 20 years. He went from being 68 to like whatever 20 plus 68— no, I'm just kidding. I did learn that much in high school. But he got old. He couldn't use a phone. He couldn't use a computer. All of a sudden, nothing was accessible to him. He couldn't go to— he couldn't go to the bank. He couldn't do anything. He became someone like— I was like, you have to go into an, you know, an aging person facility, a nursing home. He's 68 years old. So I did. I put him in a nursing home. I gave him Alexa. He was the coolest kid, by the way, at school. School, at the nursing home. Everyone wanted to be in his room because everything was voice activated. But the interesting thing about that was that he didn't do well there because he was too young to be there. So I got very focused on how to use the technology we're talking about today to not just serve the affluent, to not just serve the 1% of the 1% who are going to buy a device you can talk to in your kitchen, but how do we repurpose this technology? How do we think about that technology in a different way? So in that email that Jeff Bezos sent out, he said to all of us, you're going to join a team that's going to build the Star Trek computer. Do you know what I mean when I say that? The Star Trek computer. What that means is that you can ask for tea and tea will be made for you, or you could ask for anything and it'll just magically appear. And I was excited about that. But the reality was, is that today, the technology allows you to do exactly that. You can do literally anything that you can write a piece of software for, you can now call with your voice. The problem was developers didn't know how to make the connection. The person building the solution didn't know how to use the technology that's available. So an opportunity was created. However, I've worked on some of the largest AI models in the world, Alexa being one of them. We're now in over 100 million homes. It has, I think we're at 2 billion devices worldwide that have Alexa enabled on them. That's crazy stuff. That's very large numbers, right? B billion, like very large numbers. And I spent a bunch of time asking hard questions. And the reason I ask these hard questions was because one day I got off a stage like this one and a woman came up to me and she said, "Why would you name a device after a girl? Why would you do that?" And I was like, "I don't know. I didn't— that wasn't my response. I didn't do that. I'm just the messenger." I used to say that all the time. I don't say that anymore. I'm just the messenger. And this woman who was a mom She exposed me to a whole world of hurt people that I wasn't aware of. And it kind of comes to the heart of what we're going to talk about all day today, is that even though, like, I'm going around having a good time, I'm getting awards, I'm talking about AI, I'm building the largest AI models in the world, meanwhile, this woman represents hundreds of thousands of people that are that care about the fact that Alexa was named Alexa. And the reason they care is because, just interestingly enough, a bunch of young women, girls, moms, I guess, moms named their daughters Alexa. Is there any Alexas in the house? 'Cause you feel probably like I'm talking all to you. Okay, not surprising, they may have changed their name by now. But Alexa, there was a bunch about 7 years before Alexa was in its highest sales ever. This was around 2015. We were selling a million devices a year like crazy. And fastest growing AI technology on the planet at that point. But there was this little group of women, little, 100,000, that had named their daughter Alexa 5 years, 6 years before. This year that Alexa became part of almost all of our households as you saw when we all raised our hands, those kids were going into 1st and 2nd grade. Do you know anything about 1st and 2nd grade? Do you remember? It's like when the mean gene kicks in, the bully gene, the, like, I'm gonna make you feel bad about yourself. I don't know why, it doesn't happen to all of us. I was the nice kid. I think you all look like nice kids. But you know it when you see it, right? When you see a person, you're like, oh yeah, that's not, we're not gonna hang out with that person. But it happens right at this age, and I used to think it happened later, but now I have data. Ew. So now I have data that shows me that actually this happens quite a bit at this age, 6 to 7. And here's what ended up happening for these young women in droves, so much that a class-action lawsuit is being planned against Amazon as a result. These young girls would go to school and someone— we won't name names— will come by and knock their books out of their hand, or knock their pencil off their desk, or, you know, drop their glue, and they will command that girl like a robot to pick it up. Alexa, pick up my pencil. Alexa, pick up your stuff. Alexa, go to the bathroom. All day long. Like, where they— of course, bullies always think they're funny, but It got to the point where these little girls didn't want to go to school anymore. They not only did they not want to go to school, they started asking their parents to change their name. How sad is that? That's sad to not be able to own your identity because some group of data scientists were like, "Yeah, Alexa is really good because of the lexicon and canonical phrase that does not cause false wakes. We should use that." Did any of us think about the little girl who would get bullied in second grade? No, we didn't even ask the question. It didn't even occur to us the kids would even be involved. As a matter of fact, when we started on Alexa, we never thought anyone would buy more than one. We built it for one, and when you started buying two, all of our systems crashed because they didn't— we didn't know how to disambiguate two devices in in the same house. So yeah, no, we never even thought about the name. There is a cool story about the name though. Alexa is born— the name was actually a bunch of reasons. It is a great lexicon-based name, which means it's got these consonants and vowels that don't show up often in the human language unless it's your name. Um, so that was a good reason. But the other reason why is it was named after Alexandria, the largest library in the world, and it was meant to be this noble thing. And that's just one, that's just one of the very sad stories that I get to tell. Not all today, don't worry. But another example, and the only reason why I think that happened, and even to this day I can't do anything about that, right? I don't work for Amazon anymore, I can't do anything about it, but I can take responsibility responsibility, right? And the only thing I can do from my very humble perspective and taking responsibility is to let you know that the fact that I didn't have anyone on my team raise their hand and go, "Do we think about all the kids that are going to be named Alexa?" We didn't have a mom. I mean, I was a mom, but I didn't think about that. I was focused on my son. I didn't have a daughter yet. So it's just really interesting, right, how, how that happens. Anyway, I'll share with you a couple other stories about this, but though I've gotten a lot of awards and it's been a very good ride, it's been pretty painful too to see how my just very simple oversight, very simple lack of critical thinking, can actually cause pain to a significant group of people. And this only happens because we're leveraging artificial intelligence and automating scale, getting what we're doing out to hundreds hundreds of millions, billions of people at one time. And you're going to hear lots of examples of this over the next, um, you know, throughout the day, uh, all across different domains, which will be really fun, different industries. But I want to start with something that I've worked for some of the best— okay, some of the coolest— no, um, most rich CEOs on the planet. And here's the good news: they all referenced this quote. Jeff Bezos, Satya Nadella, it's pretty cool. He's Microsoft CEO, if you don't know. Arvind at IBM. Jim Whitehurst used to run Red Hat, an open source company now bought by IBM. But all of them mentioned this. Jeff Bezos actually tweeted this out and said, "I have my kids read this. It's on our fridge." And he took a picture of this quote on the fridge. It's a pretty interesting story, so I want to share it with you. I'm gonna read it. I know you're not supposed to read it, but I'm gonna read it because it has a good story. Okay, here we go. So what is success? To laugh often and much, to win the respect of intelligent people and the affection of children, to earn the appreciation of honest critics and endure the betrayal of false friends, to appreciate beauty, to find the best and others, to leave the world a bit better, whether by a healthy child, a garden patch, or a redeemed social condition. To know even one life has breathed easier because you have lived, this is to have succeeded. This makes me happy that the biggest tech moguls in the world are even aware that a quote like this exists, let alone trying to teach their children that the barometer for success is different. But that's not the only story I want to share with you here. You'll notice Ralph Waldo Emerson. Are you familiar with him? If not, he's a cool guy, like poet laureate, American metaphysician, really nice author. You can Google him, he's pretty popular, very popular. He's quoted as being the author of this quote, and when I got off one of the stages that I was on, a man came up to me and he's like, "You should research that a little bit. Just do, you know, go to page 2 of Google." And so I did. I'm coachable, right? I'm like, "Hey, if you say, okay, I'll go look." And I go and find out that the author actually isn't Ralph Waldo Emerson, the cool-looking white guy who does poetry and stuff. It actually was this lesser-known woman, Bessie Anderson Stanley. And at the beginning, she was well attributed, which is why she's on Google. She was well attributed at the beginning, but as Ralph Waldo Emerson got more and more famous using this quote, her attribution was lost. The lineage of her ideas were lost in the fanfare of Ralph Waldo Emerson. And so now I, of course, I give her credit, but this is one of the best AI stories I know, is that as we move into a world— you're going to hear a lot about how do you create visualizations of data? How do you do good storytelling with data? How do you use AI to tell a good story? Great, you can tell an amazing story, but don't forget the lineage. That is part of ethical AI. How do we remember the Bessie Anderson Stanley? How do we make sure that as we start telling this story and it gets handed down from 1 billion devices to 2 billion devices How do we make sure that the data that we collected, who gave it to us, the rights they gave us, how long we can have it, how are they getting paid, that none of that gets lost as we get excited about its velocity? I will tell you, no one thought Alexa would do what it ended up doing, except Jeff Bezos maybe. I'm sure he thought it would. But we all, right, we were kind of flying by the seat of our pants. We were kind of like, oh my gosh, this is amazing, I can't believe— I just got 100,000 users, oh my gosh, it's at a million, it's a million five. I was an indie dev, like I had no idea what we were— we didn't ask the right questions. And so now I take it upon myself to make sure that one of the measurements for success is that we preserve the lineage of data, we preserve where your data comes from. It's why blockchain is so cool and interesting to me now. Right? Within 5, 10 years, we'll see blockchain provide this preservation tool. We're not there yet, we're still figuring it out. But that is what it does. It creates a smart ledger. It allows Bessie to never lose credit for the song, for the art, for the content, for the code. You get to preserve the lineage of who owns the data. And that is one of my biggest ethical responsibilities. Abilities as a data scientist. So I have learned a few things along the way. Of course, I'm going to share with you, um, there's over 100 AI models today that is accessible to you through a few lines of code. How many of you would call yourself a coder? Be proud, it's a good profession. All right, a few of you. That's good, I like it. Um, the reality is, I mean, almost everyone I know thinks programmatically about something. And what that means is that code is usually a process, a procedure, a way of doing things. And yes, you do have to learn a language, but that's only temporary. We are now moving into the world of low-code and no-code. And what that means is that you don't actually have to learn a programming language to leverage the technology I'm gonna show you in the next 20 minutes or so. So that's the good news. You don't have to leverage code, like learn Java, Node.js, TypeScript. You don't have to know all of that. You can actually build it in a browser. And then once you have got a proof of concept, give it to somebody else and they can code it, or you can have the code generated by AI, which is the world we're moving towards. So the code part I'm less worried about. I'm glad we've got some coders in here because you're extremely useful knowledge and excitement. That's good. But it's not necessary to do what we did and to build. However, with these hundreds of models that are available today, hundreds of them. So there's models that can translate on the fly. If you want to build an app, a mobile app, and in that mobile app you want to be able to speak in 90 languages, you can do it with 3 lines of code. And you, again, you don't necessarily have to know those 3 lines of code. You just have to know that they exist and partner up with one of those people. I hope you looked around at the raised hands people. Get one of them and link it up. So we've got more capability than ever, beyond generations of capability, right? So like my parents, their parents, the parents before them. Like, we've got more capability than all those generations combined, which is incredible. However, I will see if we're all like friends or something right now, but there's a quote that I often use, and maybe you all know it: "With great power comes—" Oh, phew! Okay, we're friends now. Okay, I can relax. Yes! Anyone remember or know who said it, or who they think said it? Yeah, Uncle Ben! That's right, Spider-Man! Good, okay, we're all on the same page. Yes. So, and Uncle Ben had something to say for the, you know, other people. Voltaire also said it. I know, I looked it up. It's fine. There are other people who have said this phrase, but Uncle Ben made it famous, kind of like Ralph Waldo Emerson. But with great power comes great responsibility. What that means is that you are now going to be able to accelerate the development of any idea that you have. If you have an idea for a mobile app, you can build it in a couple of days without learning a programming language. If you want to learn to program, you could hop on freecodecamp.org and learn to code in about 6 weeks on your own, or in 100 days on Twitter. Like, it is more accessible than ever. So as soon as you start doing this though, I'm going to encourage you to start asking questions. Not just— and my friend Renee will come up and talk more about this later, but it's not just about who will you serve, right? It's not just about Jeff Bezos going, "We are gonna make the affluent so much happier in their kitchen." That was the original goal. Make the affluent, rich people, happy kitchen because they will be able to set a timer without touching it. They'll be able to play music while they cook. It'll be amazing. But our job, our job is to think differently. Our job is to think, Okay, yeah, cool, I get the kitchen, but what about the classroom? What about the nursing home? What about the speech pathologist? What about healthcare? What about retail? Our job is to think more deeply about who we can serve. And then, it's also to think more deeply about who we'll hurt. And this is hard because almost always when we're building something, Alexa, Microsoft, every AI model I've built, I'm always thinking I'm doing the right thing. It's always my intention to do the right thing. And yet still people come up to me and go, by the way, that thing you built hurt me. What? I didn't know that. So how do we stop that? We stop it by asking better questions and by following a few simple principles. So the first principle, um, is around— well, I'll give you some context first. So as you know, I worked at Alexa. That was fun. 4 years. I call it, uh, I got my MBA at Amazon. I didn't really, but I felt like it. I learned a lot about everything, um, being in that startup and growing it to the size that we grew it. I then went to Microsoft and built, uh, and was part of a team that created 17 research projects. Fun little projects. I'm going to show you them now. They're teenagers now, um, but when I was with them, they were little babies. And we took these research projects and we productized them, which means we put them in a situation where people could call them. Anyone could call them. I called it AI for everyone. I used to say that, AI for everyone, everyone can use it. And then I realized I actually don't mean everyone. I mean only nice, good people, right? I don't really want bad actors to use this technology to create systems that will destroy healthcare. Really don't want that. But it's interesting, these are the ethical considerations of democratization. When you create a democratized framework, when you decide to make something available to everyone, do you really mean it? And is it okay if you don't? There's no right answer. Microsoft was one of the first to take a stand on actually evaluating use cases for its technology, and it will shut you down if it's not confident you're doing AI for good. I think that's awesome. Some people would call that censorship, right? It just depends on who the person is, what they're doing. Um, it depends on where you come from, what your history is. But in all of these massive projects, I learned some pretty key principles, and I'm going to share these with you, and I think you'll see them as you move through your career. Hopefully it'll get better because it's pretty dismal right now. The first one is inclusive engineering, and this is known— rather than— you probably have heard the term inclusion, right? You've heard this term, it's pretty popular. Inclusion, which often when we say it, we typically think brown people, girls, women, um, but I actually believe in something a little bit more, I don't know, maybe elaborate than that. I call it a symphony of talent or a symphony of data. I'm looking for, when I, for example, built the technical management team that created 4 languages on Alexa, when I did that work for them, I realized that I needed to have a mom, I needed to have a single dad, I needed to have someone— someone who went to school, someone who didn't go to school, someone who went to Ivy League, someone who went to state. I needed someone who represented a cat lover and a dog lover, a kid lover. I wanted someone who represented teachers, someone who represented people like me who cared for their parents, their aging parents. I needed to create a symphony of life experience so that when we ask the question, "Who could this hurt?" there would be people in the room who are like, "Have we thought about grandparents? Have we thought about kids with Down syndrome? Have we thought about, you know, any divergent community?" And so that's number one, is I have to have people in the room that actually know to ask the question, that have lived a life experience that says, "I want to protect the interests of everyone in my life." And that means, "Oh, guess what? I never asked you where you went to school specifically or what you studied. Some of my best favorite data people, data scientists, data engineers, actually started off in a totally different field. One of them is a record producer turned data engineer. One of them is an artist turned, uh, like an oil painting— I don't even know what they're called— but oil painting artist person. Another one was a sociologist. I had an endocrinologist. Right, these people, yes, maybe technically inclined for some of those fields, but for the most part had nothing to do with AI, had everything to do with humanity. And that was really the gap we were, we were facing. So inclusive engineering doesn't just mean though having a table full of people with unique thoughts, a symphony of talent. It also means a symphony of data. You have to actually get a diverse data set. To give you an example, one day I had a woman come up to me and she asked— this is early, like maybe 20— doesn't matter, but early in the days of Alexa. She's like, "Why does Alexa listen to my husband more than it listens to me?" And like marital struggles ensue when Alexa listens to one spouse over another. Or if I am screaming at Alexa and my husband walks in and says one word, like the same thing, and it like instantly does what— that's not okay. Like, bad things happen. So I immediately was like, I know what's wrong. Does anyone know why that happened? What was that? Yes, data. We didn't train it on women. Ew. Why? Here is the sneaky part. Did we want to train it with female data? Absolutely. Guess what we did to collect data? We sent out an email to everyone in Amazon— lots and lots of people— and we asked them in their spare time to supply data. Guess who the only people were that could come in early, stay late, use their lunch hour? Guess what? They happened to all be a shared demographic. Not because our whole company was that demographic— it's a lot— but No, we had a relatively good level of diversity in the email that went out, but we framed the participation in that data collection in a way that limited the diversity of the data inherently. Oh, we didn't even know to look. We were like, "We asked everybody. It's not our fault that only these people showed up." We didn't even think to ask about fault. We were just like, "Awesome, we got 250,000 people." to give us data. That's our dataset. Let's go! It seemed great until the customer tells us we missed it. I always tell people, you do not want to find out about the good or bad part of your product on subreddits. It's not the way to go. Like, that's the worst place to find out that your product stinks or hurts people. At least it exists though, so use it, because at least it exists and I know to go there and I can go, oh my gosh, we We messed up. Another one, oh my gosh, how many of you know who Marshawn Lynch is? He's a football player. There's a few of you. So pretend you know about football, and Marshawn Lynch, he's a character, like a character. He's on Skittles commercials. I think he, I don't know, got sponsored by Skittles at one point. He's on a lot of different commercials today. I don't know if he's retired. It changes all the time. But back then, he was on Seattle Seahawks football team, and they had just, I think, won then lost the Super Bowl or something. And Alexa wasn't working for him. So he calls up Jeff Bezos. I don't know exactly. Someone asked me once, like, does he have his number? I don't know. All I know is Jeff Bezos came to us and said, Marshawn Lynch says that Alexa is broken. So what they did is they asked the whole room. There was, I don't know, 60 of us, and I I'm like, who knows Marshawn Lynch? Everyone in the room raised their hands except a few of us. I was one who was like, I don't— like some of you in here, I don't watch football. Anyway, they were like, great, Noelle, you're going, right? Because they didn't want someone, I guess, who's gonna be like, oh my gosh, can I take a picture with you? I could care less about the football. Did I say that? I love football. If— anyway, if you're a football person, I did watch the Packers game. So the whole time I'm there, I'm, you know, trying to understand what he's doing. So I ask him, okay, just call Alexa right now, just call so I can see what's wrong. And he's like, Alexis. And I was like, oh, all right, we can fix this. I was like, no problem. Okay, Marshawn, it's Alexa. And he's like, okay, Alexis. And I was like, no, no, it's Alexa. And he's like, "Alexis." I know, that's what I'm saying. And I was like, "Roger that, man. Okay, cool, good. I will get this fixed right away." Guess what I did? I took that data back to my bunches of data scientists and I said, "He's saying 'Alexis.'" There's actually a whole SNL skit on this. He said, "Alexis." They trained— I took recordings of his voice. They trained Alexa on his pronunciation. And today, if you say "Alexis" to your device, it will work. That's how data works. How crazy is that? This one-off thing, and who knows how many people who were saying "Alexis" and it never worked for, it started working for. But like, is that the way to build a product for 6 billion people? I don't know. We don't know. When we're building a product for the world, when you're building solutions you hope will touch everyone, the only The only thing you can do is have a mindset, an openness that says, "All right, if somebody says it doesn't work, I'm gonna run on over, find out why it doesn't work, and then fix it." 7 years later, we've made hundreds of thousands of these little fixes, and now Alexa kind of works okay. So it gives us an opportunity to just kind of see that nothing— I mean, this is Alexa. A lot of people think Alexa was a well-funded project. It wasn't. It was like a startup. We were fighting for our existence every day. We were in the red from forever. In the red, meaning, right, like we didn't make money for the company. I still don't know if Alexa makes money for the company. But it was a dream, a dream of Mr. Jeff Bezos. Now he's on to other dreams, rockets and stuff. But at that time, that was it. But the reason that Alexa is the way it is now is because we are very inclusive. Inclusive in deciding, right, and figuring out what datasets do we need to make Alexa better. And that dataset could be from one person who represents a constituency. Today there is a model being built, a speech model, that is specific to kids who have— you could actually Google it, I think an article recently came out— for kids who have special needs, who have cognitive and speech delays. And because they speak differently, and there's actually a machine learning model that's being created to help Alexa understand them. I actually loved Alexa for my son because it forced him to repeat what he was saying over and over and over again until Alexa understood him, and that actually improved his speech. So he didn't need a special model, he didn't need to be accommodated in that way. Some do, but he actually benefited from having to say it. And here's the kicker: when he he would say it wrong, Alexa never got mad, never got frustrated, never got like, "It's Tuesday, Max." Like, he never said it like that. I say the same thing about my dad. Like, oh, you know, my dad has, he's like Dory, Finding Dory, do you know it? Or 50 First Dates. Anyway, he's not as bad as 10 Minute Tom, but it's pretty bad. Every morning, he doesn't remember what he did the day before. So he'll be like, asking me the same questions. And now I just tell him to ask Alexa, because Alexa never gets frustrated with the 15th time he's asked what day it is, what county we live in, um, the numbers for coronavirus. During Corona, like, he asked me that question every day. How many new cases? I don't know. You're a data person, right? Yes. Ask Alexa. And Alexa every time would, with a smile, answer the same question over and over again. And it created a level of companionship relationship for both my son and my dad of a persona that wasn't frustrated. And as caregivers, let's face it, we're human. Sometimes we get frustrated. And it actually helped, I think, improve their growth. So the second to last thing I want to talk to you about is the multimodal development. This is the world we're moving into. And what it means is that we want to not just think— when I went to Amazon, Jeff Bezos Bezos was always saying, voice first, voice everywhere, voice everything. And then he changed it to Alexa first, Alexa everywhere, Alexa everything. He wanted, right, dominion over all things voice activated. What I found out though is that voice is great, but it's immediately— how many of you are familiar with Clubhouse? Do you know that app, right? It's an audio app. Immediately Clubhouse alienated everyone who couldn't hear, right? Or speak, right? And so you instantly, when you say anything, if you use the word all, none, everywhere, everything, you instantly become exclusive, which is okay maybe at first. But I'm a believer that you want to open the funnel, open the doors to your software as big as possible. And the way to do that is through multimodal development. What that means is that different modalities for different people, all pointing to the same tech. So the first one, right, is of course, yes, I think if you write technology, you should make it so that I can talk to it. Why do I have to use my phone or my computer if I could just use my voice? I would way prefer it. And study shows about 78% of the, at least, American people would agree. They'd rather just talk to their tech. And that's the world we're moving tool too. But I might also want to use gesture control, or maybe I want to use a typical device. So I went to National Public Radio. If you know it, they have like cool podcasts and stuff. But I went to National Public Radio and I became their head of engineering, and they had a multimodal approach which I loved. They had a web app team, they had a mobile app team, 2 of them, one on Android, one on iOS. They had a voice team and soon to be voice on Google and voice on Alexa. And now, right, they're thinking about the metaverse and how do we create an experiential community as part of our product. How do you go in and listen to a podcast, for example, by walking into the metaverse, sitting in an audience just like this, but it's virtual, and be able to talk to some of the people that you've been listening to for years. So this is pretty cool, but the benefit is, is that the funnel of people using your technology widens. You get more customers, and more customers means more revenue. And almost everybody likes that if you're running a business. So multimodal does one of two things: it helps you increase accessibility for your technology, but then two, it actually also improves the productivity and profitability of any effort you take on. So multimodal development is a huge part of what I have found made solutions successful. And of course, today is all about ethics. And I, in my professional life, I have to be very careful about ethics. You don't here, so I'm gonna encourage you to dive on into, like, have you ever seen the trolley video from The Good Place? Yes, some of you have. It's super graphic, like trigger warning, super graphic. I did not know and I was intimidated by the amount of gore. But for those of you interested, there's a video, Trolley video, it's an ethics video from The Good Place on YouTube, very popular, but it basically talks through the scenario of let's say autonomous vehicles, right? And this is a true story. Autonomous vehicles, we won't call we'll call them Beslas, right? These Beslas were being created hypothetically, and the computer vision model had to make a decision when approached with the opportunity to collide with something, right? A small dog or a human. Anyone care to guess what it chose? Right, sounds like dog. So, like, that's a simple kind of mathematical equation, right? Small dog, not human. Human, okay, we should probably save the human. However, what happens if it's your dog, right? Some people, when asked this in a study, actually hesitated on the human when it came to their dogs. Like, I'm not gonna judge them, go you know, their family. Dogs are people too, I guess. But how interesting is that, right? When asked if someone— if you ask the driver, what would you rather me do? If it's their dog, their ethics change. And that's what the trolley video also demonstrates. If it's personal to you— that's why all of my technology choices have been around my son and my dad. That, that's the ethics that drives my choices. So interestingly enough, we have to kind of figure out how we acknowledge the ethics in any conversation, in any decision to make software, but also how do we take responsibility for that, right? So the way that Besla hypothetically decided to take responsibility— you can imagine them having this conversation is that they said to themselves, "Alright, what's the most important thing? The most important thing is the driver, so we will do whatever it takes to preserve the driver." In this case, it worked out because if you hit a human, it's much more likely something bad will happen to you and your vehicle than if you hit a small dog. So it worked out, but that's not always the case. Deers, humans, anyway, but they decided to model, build the machine learning model, the predictive model against thousands of variables, right? There's lots of variables, but one of the variables was on the security of the driver, and that was an ethical decision that the company had to make. And those are decisions that you'll have to make. I will say I'm a little bit worried today about companies that aren't even having ethical conversations. Most companies are not even asking the question about the dog and the human. They don't realize the impact that their technology will have on humanity, and that's why these discussions in this room, in your small groups, are going to be so important. So finally, today is all about— I'm hoping— learning by doing. We have a bunch of different breakout sessions, um, I think, okay, we have a bunch of different breakout sessions that are happening, and you'll get a chance to get some practical experience doing this. I'm going to be leading one. There's a really cool one coming up, but in the afternoon when we break out, that's going to be what we, we get to do. We're going to have a chance to actually put, not necessarily fingers to a keyboard, but definitely practical applications. So I want to bring up I want somebody to say something. Okay, I'm gonna bring up Joy.