 Actually, I guess I was more annoyed that you made it to me, so I was very frustrated, you know? Like, you guys know how it is to learn code programming on stuff for a couple of months and debug. Nobody told me when I started to code that it's more about debugging than actually making use of it. So I know it's late afternoon and you've been hearing people talk for the whole past days, and before I start to tell you about my work, I'm still finding a way, nice to meet you. I wanted to do a quick exercise and I'm going to talk a lot about kids and play, and I think that playing is like riding a bicycle. Like, we never forget to do it, we just need to remember. So we're going to do like a very peak energizer where you stand up and you're going to play rock paper scissors with the person next to you, okay? And I can put some music to help you get in the mood. Okay, you guys know how rock paper scissors works, right? Rock, paper, scissors. All right, that was fun, right? So if we have a time into a championship where every person that wins goes to the next person that wins and goes to the next person that wins, and at the end we're only going to have a tool where I'm going to be the rock paper scissors master. Yeah, so you got into it immediately and you showed me what I really, really believed before that we all are still kids inside and love to play and that playing is one of the best ways of learning and engaging with each other. So that's a lot of what my work is about and my research is about. So, listen to me, I'm working on democratizing like who creates with AI, who learns about AI, and I'm going to tell you about that in a second, but I think it's important when you have people coming here to tell you their opinions and ideas and take on the world, understand where those opinions and ideas are coming from. So I've been working with young people for the past 10 years, and these are some of the projects that I started. So Hackademia is the first one. It's a not-for-profit organization which the name is very coming. It's about hacking with kids and hacking academia. And the reason I started this is because I really wanted to enable young people to become creators of technology and not only consumers. And I really think that if you start to build something when you're seven and take apart your radio or car or whatever, and then go into like, oh, I love video games. What if I can add my own add-on or change something? That's what really is leading these kids to get into tech later. And not only that, it's also like an attitude towards life where you have the agency to change the things around you and the things that you're using and you're not expecting always other people to do it for you. So you ask me, kids ask us about this question all the time. And I love, I think Marty also asked like why, why, right? Like a team needs to have a goal. Why do you do this to find out? Well, the truth is that I come from like a very small village in Transylvania in Romania. I actually have an image of the map and then I was like, doesn't matter if I come in because you won't have the context. But yeah, Romania is Eastern Europe. I come from these like really like small, like I think it's like 6,000 people village in the mountains. And I grew up not having, like I got my first computer when I was like 17. I actually build it. But like when I was like young, I didn't have access to these technologies. Like I learned much later in life. And constantly, you know, as I got the opportunity to go to programs like Singularity University or NASA campus, to MIT, to you name it, I was always thinking like, oh my God, if I would have learned about robots, nanotechnology, synthetic biology, when I was like six, imagine all the things that I could have done by now. So when I started Hacidemia, I really thought about this. I thought that if we can translate some of our biggest ideas and biggest problems and biggest, you know, like questions in terms of technology in ways in which kids can understand and not only understand but actually create with, then we are really preparing them for the future. And that's what education should be about, right? That's what a 21st century education should be about. So one of my first workshops, this is in 2000. I'm gonna date myself now. This is in 2015, in Berlin, one of the first Hacidemia workshops. She is 2014. And this girl, can you guys guess how old she is? Can I get some more? Four. Eight. Eight. Five. Five more. Six. Seven. She's six. Yes, she's six. But I like that you guys think she's four. Yeah, I guess if you could read it out from here. And why is she doing, can you guys tell what is she doing? She's soldering. Can we allow six years old to soldering in our old school? Why not? Because they can work themselves. Okay, they can work themselves for this film at home as well, right? Like, who else, you know, like, we say these things because they were told to us, like maybe our teachers or our parents said the same things. Like, oh, that's dangerous for you. Or you shouldn't do this because I didn't do it when I was your age. So it's like very interesting that the same people that are trying to change educational systems and the way we learn are people who are products of a certain system, right? So they aren't always questioned. So I can tell you that a six years old can solder with no problems without burning herself, right? Moreover, she spent, she was doing this TVB gone project. I don't know if you guys know it from each of them. It's like an open hardware project where you can make a universal remote control and turn off any screen. So our dream was to build a lot of these good kids and turn off all the TVs, see pretty. She spent three and a half hours soldering her universal remote control and her parents were terrified. They were looking at her, they were looking at me. She didn't want to take a break. She didn't want water. She was like in the flow, in the zone. And two days later, her dad calls me like, Stefania, where can I find a soldering iron? Like my daughter, we've not stopped talking about this. And I think it's really important to mention this. On one hand, because I think kids can do much more than we give them credit for. And too far because I think that if we constantly protect them and try to kind of, you know, domesticify and like really dumb down everything that we explain to them or give them access to, we're not really comparing them for the future. And we're open like, I don't know if you guys know, but today it's International Women's Day, yeah. And I think it's very important to not only give kids access to these things, but actually encourage little girls and women to get their hands on and try dangerous things, right? So that's what I made my life mission to give up. And I think, you know, after having worked with kids, who are literally more than 40 countries and taking like, you don't need to necessarily have like expensive sensors or, yes, if you have an Arduino school, but in certain places, you can't even afford that. We actually had, this was the first maker fair in Nigeria in Lagos, 2012. And this kid built this cream entirely from recycled materials. And he was using syringes and had like super complex mechanisms for controlling the cream. And then, moreover, like it's not just about the kids, we invite the parents in as well. Because there's only as much as we can do in as part of our workshops or as part of our programs, it's actually super important for parents to pick this up and continue it at all. And pay attention to what their kids are passionate about, right, and ask questions. So my experience in working with parents has been diverse, like in terms of having like really helicopter parents, or like, okay, you do this, he's not learning enough, you should try this. Or having parents who are like super laid back and just want to drop their kids at a workshop and come back and pick them up. So like, we really like have different strategies to really bring people in. And you guys are in tech, so if you probably will have kids and they already have kids and go with a workshop with your children, you will not be so terrified of like learning how to make a game with your kid or programming a robot. But for people who are not in tech, it's actually up to us to design tools and activities and programs to actually bring them in as well. Because I believe coding and machine learning is part of, it's a new literacy. So even if you don't work in tech, you still need to know a little bit about it. And actually training the kids, it's a great way of training the families and the parents as well. So yeah, so basically after doing all these things, like I was working, doing workshops, training people to start their own workshops, everything we do is open source, it's online. I reached a point where I realized this is how much what I did can grow in a grassroots way. And I should also say that we had like the Robin Hood business model where kids could afford to pay for workshops they would. And that would allow us to give scholarship free access for the kids that couldn't. So after doing all this work, I decided aside from me also to go back to school. So I really wanted to go to Media Lab, which is this like, how many of you have heard about MIT Media Lab? Okay, I'm full, okay I'll explain why this. So MIT Media Lab has been created 32 years ago as part of MIT and it was the first interdisciplinary like research center of MIT. So people that do music and people that do robotics and people who do prosthetics would all be in the same place and work with each other. It was created by D-Class, Negro 40, Simon Popper, Marvin is the amount of other amazing, amazing researchers. I knew I wanted to go here, but I knew these people only accept students that have crazy ideas but can also implement them. So the reason I went to Flatter and actually applied for a scholarship to be able to go to Flatter is because I knew I need to become a much better developer like full-stack. So I can show that if I have an idea or a research question, I can very quickly implement it and translate my ideas into prosthetics because otherwise I wouldn't get accepted. So it was all part of a really big long-term plan, which somehow like crazy it worked. I should also say, this is part of the conversation I had with some of you yesterday that it took me a very long time just to decide to apply because as a woman, I always thought that I need to learn so many things and be so prepared and have all these curriculum shows before I can apply. And it turns out that these people actually, like the lab I worked with, knew of my work with the kids from before and they're like, why didn't you apply earlier? Like, and it took me so long because I was like, I need to learn this, I need to prepare this, I need to do this project. So my advice for the woman in the room, always apply. Even if you're not ready, even if you don't think you're gonna get in, like there's nothing to lose, always apply. You're gonna learn a lot in the process and I think very often we hold ourselves up to very high standards and it doesn't always work for our advantage. So when I join, I join this group and the name exemplifies directly what the group does. So the group is called Lifelong Kindergarten and it's the same group that created Scratch. How many of you have heard about Scratch? Yeah, so Scratch is like the largest online community for kids to learn how to code. It's more than 10 million kids. It's translated in 50 plus languages and it's a group of loving people, like extremely passionate and dedicated people who really care about creating learning for kids. So the poor, like part of this group is from Professor Rachel Resting, who runs the group. It's what he calls the Four Peas. So the Four Peas stand for Project Passion Play and Peers. And I think this applies for the kids in the Scratch but it also applies for what you do during school or for anyone who wants to learn how to learn and in general how to grow, right? So we learn best when we do a project, project-based learning. And we learn best when we're working on things that we're passionate about, right? And we learn best like we did earlier when it's fun because if it's not fun, then we're not gonna spend a lot of time doing it and we're social animals. No matter how much technology we're putting into our lives, we're always gonna need people in peers and communities. So whenever I'm deciding what I'm gonna be working on or how to design a project or my work, I'm always going back to this, right? We have like this focus on projects, project-based learning. Does it allow people to be playful and passionate about what they do? And does it allow for a community to grow, right? And for peers to learn from each other. So yeah, that was like what was at the base of Scratch. This is how Scratch looks like. Just how it's growing. Actually, I'm searching for their latest 3.0 and I'm gonna let some kids explain to you how Scratch works. I was searching for interactive programming language on the internet and I came across Scratch. As soon as I shared my first project, I got interested feedback and I was like, okay, I want to continue with this. I'm a records project. It's a super simple animation. I'm still actually really proud of my first project. It opened my eyes to the possibilities that coding can present. So I've been drawing basically since preschool when I learned how to hold a crayon. But I think making art with Scratch is different because I can collaborate with other people, make something interactive that other people can use. There was a song that said Happy Happy Birthday. I put Scratch in there and it just became an instant hit on the side and inspired someone who would make their own songs and remixes. Anyone can do anything with my project and sometimes it's really funny how I change things at the end. It's just good to show how much you can actually do with all the things you have from Scratch. When you're looking at other people's projects, you're engaging on such a different level which really encourages this kind of mindset of seeing yourself and each other as makers of things and not just consumers. But I never thought that I could do the coding partly. I thought like actually it's on the grown-up and I've hired some coders for me but then I was like, hey, I can actually do this myself. It gets more complex, but once you know the basics, you can add on to your knowledge and move at your own pace. If you like solving problems, if you like drawing, if you like making music, if you're into anything, Scratch is for you. Yeah, it's open source and it's online and I know what a lot of you are thinking. It's like, this is blocks, it's not real coding. What I'm gonna tell you is that it is real coding and actually a lot of these kids, like you saw, and this goes back to the previous presentation about designers versus programmers and you could see how it provides many pathways for people who are into music, for people who love drawings to actually get into code. And I actually use Scratch also to teach teenagers. I was a teacher for Girls Who Code and I like it because when you get people started, they don't get stuck because they miss the semicolon and they can operate with concepts. They can understand what a loop is. They can understand what a function is. They can understand what are the computational thinking like primitives that operate with them and then of course once they get something working and it's fun, they're gonna switch to like Python, JavaScript, you name it and a lot of these kids did but it does provide accessible like low-force, like accessible ways for people to get into it and not get intimidated. And if you actually wanna see like insane programs that people are doing with Scratch, I highly encourage you to go to the website and just search things like OCR. You're gonna see like entire algorithms, oh, Wi-Fi is too slow. But yeah, if we type here like OCR, people implemented like entire algorithms to detect like written text or written numbers in Scratch or if you just search AI, insane amount of games and very complex programs that were created. So don't underestimate like how complex the high ceilings of this platform and see it more as a first step for people who want to get into coding. So yeah, this is a neural network for recognizing handwritten numbers implemented in Scratch. And if internet was a bit faster, I would actually show you how it works, but it's online. So you can try it. So yeah, so I work with this team and basically they were preparing to move away from Flash because Flash is dying and go to known and that would allow them to also be accessible on mobile devices and be responsive. And it also allowed them to make a lot of connection to physical work, which is my first love because I started hardware. So this is another project that came from the team. How many of you have heard of Make-A-Makey? Okay, I have to show this video. It's too much fun, is it a video? Yeah, so there's this banana piano, it's not there. So yeah, Make-A-Makey allows you anything that is conductive like banana or wet paper or aluminum foil to become like an input for your computer. So you can map any conductive object to a key and then do things like, I don't know, we made a musical room where every single object could be making a different sound. You can make games where you're jumping in a real world and a character like Mario would jump in the game. So it really allows you to connect the digital to the physical. And there are many other extensions like this that allow people to basically see the world as a construction kit, right? Anything that has an API or anything that has a firmware can be mapped and connected to any web application. So one of the first things I did was take Spotify, make an extension for that, and then have five robots that are density when it's not from Spotify. So these are like a list. If you go to scratchx.org, you're gonna see like a ton of extensions from Arduino to the International Space Station Trapper to leap motion. So these are like just like building blocks that kids can put together and make projects. And what does this have to do with AI? Well, it turns out that almost half of households in the West have a voice assistant, right? And these devices were not necessarily designed for kids, but kids end up using them and it completely shifts everything we know about child development. If a four-year-old can get on the web via voice, like that four-year-old doesn't need to read and write even, right? He can just ask any question to a lifestyle or Google Home and get an answer. And why did that four-year-old ask about God if that exists? Or why did that four-year-old ask like very tricky questions like what is love? Or you know, a real problem right now is that a lot of kids are actually doing their math for more than Alexa, right? So this is a thing, it's a thing. This is like the first generation of kids that is actually growing up with AI. So seeing this, I was like really thinking like can I actually like create extensions to allow kids to teach Alexa and program Alexa and not only use Alexa? And not only Alexa, I actually came across like other devices, how many of you know what this creepy doll is? Have you heard about it or seen it before? So this creepy doll is called my friend Kayla and it's a Bluetooth connected, it's one of the many IoT toys. It has a Bluetooth connection. And it turns out that in the rush of trying to get to kids with smart toys, companies didn't put a lot of effort into the security of these devices. So this doll got hacked, and one of you got put on the market in Germany. And it's kind of like the baby monitor is being hacked. So like a random guy would just like talk through the dolls to your tile, right? Not scary at all. And basically in Europe, this doll is back. So the European Union made regulations that like, look, if you don't have like a security permission, if you're not in performance, like GTR and other things, you cannot sell this doll in Europe. However, you can still get it on Amazon for $29. This thing, it's called Aristotle. And this was Mattel. Mattel is one of the biggest toy manufacturing company. And they dedicated an entire team. They did so many resources and effort to launch this Alexa for Kids Aristotle, which also had a camera. They were planning to launch it for Christmas. And it turns out that they did think about actually asking the parents, like, would it be okay with this device? Like actually recording what their kids are doing all the time with the camera. Or doing research and see like how this could be used or like what impact it would have. So they just were like, they announced it, right? And 15,000 parents signed a petition against it, saying like, we do not want to have this in our nursing, you know, room. Like we do not want to have this in our kids' life. So they had to put it down from like, they had to cancel the launch for its product. So this gets to show that while, you know, companies are rushing to launch these products for kids. And while things like Alexa and others are already in the homes and being used by kids, we have very, very little understanding of how are kids actually using these things? How are they perceiving them? How can we do it right? So I started to look into it and this was like the first extension I built. Let's see, there's a video. So this is like an open source robot. It's using Raspberry Pi and the parts are 3D printed. The parts that are holding the servos in place. And you can teach it by demonstration. So you can teach it how to draw and then try to do it too. And sometimes it makes mistakes, right? So you can also teach it to react to different like UI codes and program different animations into it like make it act happy, make it act like a dog, make it act sad. You can also, let me see if I can speed up through the video. You can have a collaborate with other robots. How to draw balls, this is kids favorite activity. Yeah, but this is like really kind of sets the tone differently. It's like you're teaching the robot, you're demonstrating things from the robot and the robot plays pretty well, you know? And it can get more complicated. You can teach it to play rock paper scissors with you and so on. So we actually did a study where we invited 26 families to the lab, coming to the lab, and had different stations. Like each station would have a different device like Alexa, or the Kirila Doll, or Cosmo, chat box, Google Home. And then the kids and the parents play with these devices and then after that we ask them questions. Is it smart? Is it smarter than you? Do you trust it? Is it friendly? Because we just wanted to understand how our kids receive these devices, right? And how do they play with them? And what questions do they ask? And yeah, I can show you a big video to see how this works. The age range was quite wide. The kids that were older said that Alexa was smarter, it was more smarter than they were. Alexa, what is going on? Alexa, what is going on? Alexa, what is going on? Alexa, what is going on? Alexa, what is going on? Alexa, what is going on? Alexa, what is going on? Let me show you the next one. Alexa, what is going on? This wasn't the right thing to do. I know everything that's hanging. But jewelry was like a normal person. She sounded like a normal person. And if you fall in, you actually know who you are. Hey, Google, who's your best friend? If I could get you, I'd have you. I have you. So it's endearing, but again, right? Like, of course a child like you, you ask, oh, is Google called your friend? And they're going to take to the device and ask the question, are you my friend? And the device will say, I'm your best friend. And the child is, oh, that's awesome, right? So how can we actually design devices that are a little bit more honest and transparent? And we don't want to mystify all of this intelligence, right? Where does this intelligence come from? What are machines good at? And what are people or kids good at? What is normal to a triple-morphized? Like, I get this question a lot. It's like, because I work in social robots a lot, it's like, is it OK for kids to have triple-morphized robots? And I would say that our way of making meaning of things is by a triple-morphizing, right? Like, I look at the plug, and sometimes it looks like a face, or, you know, I'm having an argument with my goomba because of the cleaner. So we do that, Amy. We do that naturally. But I definitely think that there is a somewhere learning into how much should we design these devices to replace friends or parents, or how much should these devices actually communicate to their kids why they answer something in a specific way, what they know and what they don't know, where do they get their information. So because I was very surprised that all their kids said that all the kids that were like six and up said Alexa is smarter than they are. And the kids that were younger were, like, not so sure. That was very surprising for me because, you know, there was this girl who loved slots, and she could block Alexa by asking questions about slots with three questions. But she still said Alexa is smarter than she is because it has so much information. So I'm like, how about kids, you know, define smart for machines different than smart for people. Maybe it's like a semi-optics. Maybe it's like the current they're using. Maybe, you know, I really want to see what are kids' expectations in terms of smarts for machines compared to smarts for humans. So we made another study to try to ask that question. And in this study, we actually had kids watching videos of a robot, of course, not solving bases, and then watching videos of mice solving bases, and then having to teleoperate Cosmo through a Lego maze and trying to solve the maze. And every single time they would watch the videos, we would ask them, how smart was the robot? How smart was the mouse? How smart were they at teleoperating? And the reason we picked this task is because it's very clear to see success when it's a maze, if it's solved or not. So it's easier to determine that way. And the best thing is that we actually asked the parents to do it as well. I love this picture because it's like a small chair. But they took part in the study separately. So we invited the kids and then the kids went out of the room and then the parents came in and they didn't get to talk to each other. And what was very interesting is like, first of all, I want to share something that really shocked me, like a lot of the younger kids thought the mouse was a robot. So I had a lot of kids that were, and bear in mind, this was done in Boston and they could have both Boston area population. So I think those kids are more exposed to technologies than other parts of the world. However, it was still surprising to me that a lot of the four years old and five years old had more experience like becoming new for five types of robots, but they never played with animals. And they thought the mouse in the video was a robot. So that was interesting. And then the other thing that was interesting is that even if these kids had coding experience before or technology experience before, what impacted their decisions or whose marker and the way they were describing it was actually the way their parents were talking about it. So parental attitudes and even words, like explanations would carry on to their kids, even if they did the studies separately, right? So we would see pairs where the parents would say, yeah, the robot is stupid because it was programmed. So someone gave the instructions and they gave the same thing. Or we would see another pair of a child and a parent where the parent would say, oh yeah, the robot is smart because it was programmed so it can detect like colors and edges and learn and the kids would say the same thing. So here we see two things. One that kids are actually mirroring parental, like mental models and the way they talk about these things, but also that we need new terms because right now when you have a robot or a system that can learn from machine learning, we are still saying it's programmed, right? And that's not really like the same thing as in the case where I'm just sending some instructions to the robot and the robot is only going to do that. So we actually need to come up with new terms because if we do not have the terms, we are not able to develop an understanding of how these things are working. And yeah, like our study actually showed that the younger kids didn't mirror their parents as much. So the older the kids would come, the more they would actually not only choose the same things as their parents, but also talk in the same way. And yeah, I think it's very important, this gets to show why it's important to start early in introducing kids to these things, but also why it's important to train not only the kids, but also the parents. Last, I'm sharing quite a bit of research video. This is the last reference I promise. Remember the creepy doll, right? Yeah. We actually did another study. Have you guys heard about the marshmallow challenge? Yeah? So we tried to do the marshmallow challenge type of study, but with the doll. So we actually had kids coming in, playing the game with the doll. We had three conditions. One where there was a doll, one where we had this called human confederates, so basically a graduate student, and one control condition where there was nobody. And kids would watch videos that have animations and statements like, is it okay or not okay to stand when everyone is sitting? Or is it okay or not okay to hit another child? It was like a picture with someone hitting with a pillow. And these were questions that are standard questionnaires that were used before to test conformity and moral values. And after they would watch the videos and answer the questions, they would play a game, and then they would go back and watch the videos and ask the questions again, only that this time the doll would try to make the kids change their mind, right? So the doll would say, I think it's okay. I think it's not okay. So it would try to persuade them one way or another and say to the human confederate, and in the control there was no intervention, but we just wanted to see the kids would change their mind. So what we saw is that actually when it came to moral questions, like is it okay or not okay to hit another child? The doll had a bigger influence in changing kids minds than the human confederate, right? So granted, being a graduate student, which was an adult, if it was a child, I'm curious how their results would be different than we're actually replicating, like we're continuing this study, this was a pilot study, but still like it gets to show that these devices do have an influence on kids, it can have an influence on kids. And actually like very recently there was another study published in Nature that showed that robots can have peer pressure, can insert peer pressure in kids. So if a child is in a group of robots and they're looking, I don't know if you have this optical illusion like which line is longer and the robots can convince the kids to pick the wrong line. So this gets to show why it's important, like this is especially for parents, like it's like was it important for kids to learn about machine learning? Well, we do want them to have a critical understanding of what's normal for alexa to say, what's normal for an adult to say. Yeah, is it okay for this device to actually tell me this or ask me this? So we wanted to have this critical approach and a critical understanding. And more awkward, we really want to connect to families to not look like this, right? How many of you have traveled on the plane and see like an entire family, everyone in front of their devices, or even at home? And we want connected families to look more like this, where we're actually like, where we have parents and kids learning together. So based on all these different studies, and there's many more that I didn't get the chance to share, and based on a lot of these questions and interviews with parents and kids, we built this platform called CogniBates, which is building on top of scratch that kids are already familiar with. And the purpose of this platform is to enable kids and parents to learn about machine learning and AI by playing it by doing. So before I tell you a little bit more about it, I want to show you a video of kids presenting it for themselves. Let's do the doodle. So we were programming robots. You could play rock, paper, scissors, paper, scissors into the camera, and I'll show you one of the motions. And the camera did one of the motions inside rock, paper, scissors, shit. The computer gets like better as you play the game because like we might not know everything at first, but if we try, we get better. Everyone has heard about machine-based learning or artificial intelligence, and there's no question to ask for a lot of more tech savvy and parents who can go for it. Technology's going to be a huge part of their lives, which was something in my life. It's scary for some people. It's AI, technology, they totally get it, but as a parent and as a teacher, I thought it was really important because these are skills that 21st century kids need to have. When my dad was young, he bought a car and took part to see how it worked. So we teach people that young, how these things improve the mostly program, and how it works. So yeah, so kids that are seven years old think that AI is the car of their generation, and basically the platform is online. It's open source. You guys can play with it today. You can contribute to it. I like extensions, and it has a lot of extensions, like anything from a music headband that allows you to come from robots with your mind to sentiments now this is on text, to training your own classifiers for text or vision. I prepared a lot of demos, but I'm running out of time. So here's the website. The code lab is where you can use the blocks, and the DJI is where you can train your own classifiers. So we made one that can recognize doodles, for example. So this is very quickly put together. It's a classifier that was done in under classes from the pre-draw data set. So you guys can tell me what to draw. I don't care what person to draw. Sorry, this is going to be fun. Banana. Banana. Okay, I'm going to start with banana. And then I see a line, diving boards, syringe, okay, let's try again. I don't know if I can make it better, but I'm not on the phone. Like this, like this, like this. Okay, I'll try an umbrella because I know how to make that. And then let's see. The internet is slow, but it usually, this is, this is, he's an umbrella person. But what we want to do is actually keep adding their own drawings and training their own. So they can do that already, like, for vision and for text. So if I go, this is one that is already trained, but I just wanted to show you how you can quickly train your own. So in fact, you need to get a free API key. For this one, we're using Clarify. So if I make one that is like cats versus dogs, I'm going to delete it. Let me just do a new one. And it works pretty fast when you have good connections. So I'm going to say I want to have cats and I want to have dogs. And then I'm going to add examples of cats. And I can add multiple. I have these ones. And then it's going to show me, oh, there it is. And then I'm also trying to add more of the examples to confuse it, like a cat that looks like a dog or a cat that looks like a food. And anyway, once I have 10 examples for each, I can go to my project and teach it. And that would actually give me a model that I can use in any project I want. So if I go to my code lab, then I can use that model. I'm sorry for the internet connection. Let's try. Well, this is loading. Let's try with this one because I already preloaded it. So for example, this one, I train the model to recognize when I say happy things and sad things and backhand compliments. And this was something that kids came off with, right? I didn't even know that backhand compliments exist. So the model is waiting. And if I say something like, good. Oh, he thinks he's backhanded. That's right. So you create a classifier. And then you can make backflip for your scissors or any animation you want. But this is a much more fun way for kids to actually learn how measuring works. It makes it much more interactive. And yeah, to wrap it up, because I definitely want to have time for one or two questions. I just wanted to say that this was designed with kids and for kids. So anything from the character's design to the text that we put on the blocks to the interface for this, everything was done with kids. And what we see going back to my initial point after we did long term studies, like six weeks going to schools every single day to public schools, private schools, community centers, like really across the board in terms of socio-economic status and also experience with coding, what we see across the board is that kids actually change their opinions. So they don't say that, you know, Alexa or the Robin is smarter than they are once we show them how to program them, once we show them how to train them. So it definitely has an impact and it denies devices. And we also see that I've done this in my world, so not only in West. The different cultures are more skeptical than others. Like you can see here in Romain the most skeptical kids in terms of intelligence attribution or in terms of like how much it tells the truth. So yeah, I ran out of questions of time but like I should say that this is not, it's not only my work, it's also this is my team. Most of my colleagues are still undergrads and they started working on this at 19. So if you think you're a junior developer and you cannot have a big impact, think again. And yeah, we're always looking for people to help. We have a ton of bugs and a ton of things that we want to make better. Our code is online. I'm also going to tweet these links if you don't have time to write them down now. So definitely help us make it better if you can. Or just tell us what's broken. A lot of things are broken. And yeah, I'm also teaching a class for graduate students. So we also want to promote, like not only criticize what exists but actually create alternatives. So what's going to be the leg of AI? So I'm teaching this class, the syllabus is online if you're curious. It's hacking smart voice. I'm teaching it both at ITP at NYU and at BC. So I definitely have that situation where ITP students were more like creative coders were like, you are teaching at BC. You know, people know how to code. And I'm like, yes they know how to code. And then the people at BC say, I think they know how to design. And now they're all working together. It's awesome. But yeah, I definitely think it's important to do this kind of work and you are coding in deployment and research and bring it all together because I really want to prepare our kids for the future. And I hope you guys can help with that. And thanks for your attention. Hey, my name is Erin. You talked about how children would mirror the language of their parents. Do you think there are healthy ways for parents to talk about intelligent technology with their kids? Absolutely. Thank you for the question. I think a lot of the times parents should be just honest when they don't know how it works and try to show kids how they would look it up and learn together and same applies for teachers. I think we're way past that stage where the teacher or the parent is the beholder of knowledge. And it's very likely that the kids will learn much faster and find the answer much faster but we still need the parents and the teachers to be there as facilitators and guide. I don't know if you guys have seen there's this Momo challenge on WhatsApp that causes suicides of teenagers over South America. And these are situations in which we see clearly that young adults and kids don't know the limits or what is normal for technology where we want parents and teachers to be in the loop. So in my thesis which is also online and I do get a chance to present everything I actually show lots of excerpts from interviews and what are ways in which we try to take a term a very complicated term like neural network and what are the ways of being used to talk about it with kids. And I think it's hard for parents that feel overwhelmed themselves but I think if it's being approached like hey this is something we're figuring out together right? Or this is a decision we're making together should we put this in the home or not? Like for example, Alexa speaks German now nobody puts it in the home in German right? Because they are we do not know what data is being collected who is it being shared with and what effect this has on the kids. So we're not going to put it in the home even if they give it to us for free which is why what is happening echos in the home they're pretty much given for free even if it's given to us for free it's convenient or whatever it's cool we're not going to put it in the home until we figure it out and I think that's a much more healthy approach. Hey I'm Josh I have a son who isn't old enough to read or like program yet and I'm curious if there's anything like there's a lot of evidence that suggests that reading with your kids will like encourage a level of reading or something like that I think that you've seen things that parents could do with kids but who aren't able to read yet that might be able to start encouraging like a curiosity or excitement around the students. Yeah there's thanks for your question Josh So I'm my focus group age group is like 7 to 14 but I didn't work with younger kids and the great resource is Alison Gopnik who talks about the gardener and the carpenter and different parenting styles and why we want more like gardeners and not carpenters so one big thing for very young kids is like just being present paying attention to them there's actually like this is quite sad there's like lots of recent studies that show that kids learn how to speak later because parents put their smartphones so much so they don't see their parents when they're talking to them as much right it's crazy like so at a stage where we need robots and phones and Alexa to actually bring kids and parents back together to learn how to play together to learn how to do storytelling like all these things that we grew up with that are kind of like being affected by current use of technology right so that's one thing and I can also say that you know like I think I had a big advantage of getting born when I grew up and growing up in you know was very helpful in that way because I got to like be in the nature a lot and read a lot you know like learn how to get born and have my imagination developed through that I also encourage you to give your child time to be a child right a lot of doctors must prescribe play because kids are so overbooked that you know they get depressed and then the doctor actually has to prescribe play for kids it's like that's kind of insane but with my students I didn't have the chance to share this but with the students that are designing now for kids you know they would make this one of the students wanted to make like a game for like playing like playing grounds for kids so they wanted to do like kids that's like a playing ground and I'm like why would kids want kids like let the kids go outside go to the real playground go and observe how they play and maybe you add something that makes sounds when you're using it or that amplifies what you're already doing but we do want to encourage kids to still do a lot of physical activity go outside not being in front of screens all day long and I think the fact that car permits and scrubs and all these things can work all about is going to help a lot because there are lots of games that I can do with a phone and train it like breakfast registers or when I'm running and click that day hour but like merge it much more with things that still allow kids to go outside and play yeah I don't like to answer your question and very quickly I wanted to show you guys how simple the code is so this is all like JavaScript so the extension is the sentiment extension this is basically how many lines 101 lines of code and this is taking like a pre-train fairly complicated library that allows to do sentiment analysis on text and just creating a couple of blocks that allows kids to very quickly play and take it and make games with it so in the beginning here we're just like defining what's going to be on the block like when the text is negative, when it's neutral give me the feeling of the text and then here I just have like my JavaScript functions that are making a call to the library and say give me the score for this text if it's above this number it's positive, it's below, it's negative so this is one of the most simple extensions we've made, of course it can get more complicated but yeah, if you have like a sphere or any other like weird model that you might use or anything that you would love to make an extension for I'd encourage you to have a look at this thank you