Good morning, Discover. Welcome to day two. Welcome to our special VIPs up there in, what is it, Erie Glass first class or something like that. So great to see everyone. We got an amazing day two for you. We're going to bring up a couple of colleagues from Dayforce to talk about what we're doing with AI inside of Dayforce. Our keynote this morning from Sal Khan and Khan Academy. He's going to tell you an amazing story of what he's done. First, I want to get grounded on the Dayforce promise. My name is Steve Holdridge. I'm President and COO. My role at Dayforce is pretty simple. My job is to make sure that the value that Dayforce delivers is part and parcel to your experience. From your first interaction with our brand, with our sales team, through implementation, and for your decades as a customer, that's the standard we hold us to every day. That means I get to work with thousands of you around the world, grounded in terms of what we do, hearing firsthand what's working and hearing firsthand what isn't working. We're in the middle of a transformation that's no surprise to anyone. As you know, let's be honest, transformations are never easy. It brings disruption, it brings uncertainty. It brings change. Here's the important part. You're not navigating this alone. Dayforce is your partner. Dayforce is alongside you through this. That's why it's so powerful to see so many of you here in the room today. At Dayforce, we start with a simple question. In everything we do, what truly matters? Of all the things we could do, what is first and foremost? The answer is always you, our customers. You heard David say it yesterday. It's in our DNA as a founder led organization. It infuses everything we do in terms of product design, in terms of how we service you, how we support you, how we think about the future. It's all about customer success. Making work life better isn't just a tagline. It drives how we act and what we build and holds us accountable to that. When I meet with customers, conversations always come back to a few core questions. What is Dayforce doing to make your work life easier? How are we helping you cut through the complexity of what you need to do to run your operations efficiently and effectively? How can we free you and your employees to focus on what matters most? These aren't abstract concepts. This is real, relevant to what you do every day. When payroll teams get employees paid accurately, on time, every time, consistently, that's peace of mind, that's true transformation. When HR leaders manage complexities across 10 different countries around the world with real-time visibility and confidence in the result, that's transformation. When CFOs are able to model dozens of different workforce scenarios instantly rather than hours or weeks and make business decisions based on that, that's agility. When employees can easily swap shifts to attend their daughter's play, when they can go in mobile and take a look at their benefits real-time if there's a health issue, when they can use on-demand pay to get to work and pay that they've earned every day, that's an employee experience. That's transformation for your employees. That's what we mean when we say makes work life better. These moments are the heartbeat of transformation for any company. Frankly, they're why Dayforce exists. They're behind our purpose in everything we do every day. Let's talk about the wave right now. Everywhere you go, it's in this conference. AI. There's hype, there's noise. Is it a bubble? Is it the only thing? Is it nothing? There's uncertainty. If you listen to some of the others in the market, if you listen to the tech industry, you think AI itself was the product. AI itself by itself was a solution. AI is about a flashy demo, about tech-first promises, about doing myriads of AI acquisitions and stringing them together after the fact. Here's the truth, and I know you know this from your day to day. Technology for technology's sake and AI for AI's sake, that doesn't solve your problems. You'll spend a lot of money, but that doesn't get at the heart of what you're doing at Dayforce. AI isn't bolted on, it isn't an afterthought. It's intrinsic to the core architecture of our platform. You heard yesterday, a single unified data architecture. That's what's required to support AI. That's the Dayforce difference. That is what allowed us to accelerate our roadmap, to accelerate the things that you saw yesterday. AI with Dayforce means outcomes. It means business results, means measured, quantifiable value, as David and Joe discussed and demonstrated yesterday. When applied with purpose, when applied with focus on outcomes, AI doesn't strip work of humanity, it strengthens, allows people to focus on creativity, problem solving, empathy, connection, people plus AI. That's the magic. Joe yesterday talked about the power of the plus sign. That's the world of work for all of you. That's the world of work for us. That's why at Dayforce, AI is not an afterthought. It's built into the core of our platform. It's built into how you interact with people. At Dayforce, our goal is no one needs to ask how do I use AI? Dayforce AI simply works alongside of you. It's part and parcel of what you do so you can actually focus on the work versus the technology. Customers like On The Good, Allan, and BCU Credit Union are already seeing great results from their deployments with Dayforce AI Assistant, giving employees that consumer-like experience to answer the questions they have or to anticipate the questions they haven't even asked yet. Whether it's On The Good's frontline crew texting questions on the go in their restaurants or BCU Credit Union's staff quickly finding answers without having to go to HR or someone else, that's the promise of real value from AI. The result? A more seamless employee experience. Hundreds and thousands of hours saved. It's not just about efficiency. It's about allowing your teams, HR, and the organization to focus on the things that make work life better, to focus on the more strategic things, to do the work they're meant to do. That's what we're all aspiring to, isn't it? I certainly know firsthand. In the next two speakers, you're going to hear from colleagues of mine at Dayforce, and they're going to talk a bit about our own journey inside of Dayforce and how we use these principles to apply AI to what we do every day inside the company. This moment matters because leaders everywhere are not just asking what AI can do, but how do you put into practice that really works for your people, what's really powerful. What we found at Dayforce in our journey internally in the use of AI is the importance of this partnership between HR and IT, the need to add the people plus technology together in terms of a seamless approach that only works together. That's what you're going to hear today. We're going to bring out Amy Cappellanti-Wolf, our Chief People Officer, and Carrie Rasmussen, our Chief Digital Officer. Together, they're going to talk about practical, everyday examples of what we are doing inside the company to reshape the work, the things we're running into, the lessons learned about that. Amy and Carrie, along with the rest of the company, have been at the center of this journey inside of Dayforce. How do we apply AI in everything we do? We'll show you what it looks like when you take AI from concept to everyday impact. As you listen, I invite you to think not just about the future, but about the work you're leaning into today, about the work you and your teams are leading with, and how AI can make it more impactful, more human, and more meaningful. As you hear from Amy Capelanti-Wolf and Carrie Rasmussen today, and as you hear from Sal Khan, this isn't just about technology. In fact, it's not even primarily about technology. It's about amplifying people. It's about possibility. It's about one of the speakers in our forum yesterday. How do we increase the IQ of every single employee in your organization by 30% to be more effective at what they do? The future of work is not just about moving faster, but it's about allowing people to do the work that's most important for them to do. Thank you very much. With that, I'd like to welcome Amy and Carrie to the stage. Kind of feels special with the moving chairs for us. I know they brought in the white chair. No one does that for me at home. Hey, everybody. I'm Amy Cappellanti-Wolf and I have the complete pleasure and a lot of fun being the Chief People Officer for Dayforce. I also share a lot in common with you because I'm also a customer zero. I use the Dayforce platform every day and also help influence the roadmap as well as how we use the product in our company. I'm a customer just like you. I'm alongside with you. I see you, and I'm so excited to be here to talk about what we've learned related to AI through a people lens. I'm Carrie Rasmussen. I'm the Chief Digital Officer here at Dayforce, and I get the privilege to help lead the AI transformation. Like many of you, we don't look at just the HR professional. We look across the organization and say, how do we enable everyone to participate in this AI journey that we're on? We are here to talk about AI because it's on everybody's minds. It's hyped out everywhere. There's a lot of fear, consternation, and concern. Today we really want to talk about what we've learned in our journey and also hopefully share with you some of the best practices and worst practices that we've learned as we've rolled this out. Because this is an important relationship. It's one of the things that Amy and I have really learned is how we partner on this journey. Right now, a lot of you are probably feeling the same way we are. There's a lot of hype, a lot of noise out there. There's a lot of headlines. CEOs making bold statements that are making you feel uneasy. There's industry experts that are predicting the future, that are making your employees apprehensive and nervous. Right. Can I get maybe a show of hands? How many people are feeling that way right now? Are you feeling the hype and the uneasiness? Yeah, we agree. We see you. We understand we too are there. I think today, hopefully Amy and I are going to leave you with a few nuggets and things that you can take back to help lead your AI journey at your company. There are lots of angles to look at AI. You could look at it through an ethical use lens, through a governance lens. Today we're going to focus on the people lens because we're a people-centric company like many of you are. We're going to look at it through how you engage your people to be part of the journey of AI. First, let's get grounded in some facts and feel free to take some pictures. When you see these facts, they're pretty startling. We have a couple of surveys we're relating to. One is the MIT study that 95% of generative AI pilots fail to deliver measurable business value and ROI. The primary causes aren't about the technology itself. It's about change, integration, organizational readiness, and just the pure execution. Often it's easy to blame the technology, whereas it's more about how we use the technology and how we introduce it. We also at Dayforce do annual Pulse of Talent surveys. We just completed our 16th annual survey and we also got some startling statistics. 71% of the people surveyed have not received training. They're all kind of waiting for this training. Even more importantly, 63% of those surveyed felt like AI is really critical to their career growth. Imagine no one's training me, and yet I feel it's critical, otherwise I may become irrelevant. That's part of, I think, the concern people are seeing. Carrie, I'd like to get your point of view on the dissonance because there's a people issue and there's a tech issue. How do you think about those together? Let's think about Emigu. What first? What we know first, that number, 63% feel that AI is important for their career. We feel the same way. AI is here, it's important. It's part of the future of the work. A lot of you are already using AI today, whether it's in your personal life or your professional life. Right. The thing we know is we're going to have AI assistants in our world. We envision that every role is going to have AI agents assisting them to do their jobs. One of our speakers yesterday said he would predict that there's going to be about 5 on average per role available. We also know that this is a transformational technology, which again, I think is why this partnership with Amy is so important. As we look at how AI is going to transform the way we do work, in some cases, the technology can perform more than one role, which we have to look at. How do we rewrite that job description? How do we create the role of the future? It's not just about applying technology. It's about this partnership when we think about the person. I want to give you guys a little bit. We've been on this journey together and we looked around over the last couple years and we listened to people out there. We realized there's a lot of experts saying, oh, I'm working with so and so, or there's people saying, I'm trying this technology. We learned that there really wasn't a playbook and we were going to lead, not follow. What we did is we created our first AI Leadership Forum. Of course, the first person that I tapped on the shoulder to join me was Amy, because I knew we needed the Chief People Officer because this change was a cultural change. It was an AI culture that we were creating. The second person that I brought into the mix was our Chief Marketing and Communications Officer, Eric Glass, because we knew it was important to communicate authentically and our core values, keep our core values to our employees. The fourth person we brought in was Chelsey Griggs. She's our Transformation Officer because we knew this kind of investment was around transformation. I want to give with you guys. Some of you are saying, I'm struggling. I've had a few conversations with many of you at lunch over the last couple days and you've asked me, say, Keri, how have you started this conversation? What have you done? I'm going to give you something. You guys can write this down and you're welcome to take this away and use this if you need a starter kit to start the conversation in your company. We built a three tier approach. The first tier was democratization of this technology. We wanted to get it into everyone's hands, it's your personal AI assistant and you're going to most likely deploy one of the four leaders in this industry, in the space. It's either going to be like Perplexity, Claude, which is Anthropic, OpenAI ChatGPT, or Microsoft's Copilot. At Dayforce, we have deployed ChatGPT Enterprise to our organization as our personal AI assistant. Wave two, every role is going to have an assistance agent that is helping them and it's going to come to you in the software that you own. I work with the marketing departments, sales organizations, accounting, and Amy and HR. You saw the Dayforce platform, there are agents that are coming to you to help your role be more efficient and effective. The third wave, because it's really easy for us to stay in wave one and two, I just want to be more productive. As Amy comes in, she says it's hard for me to make organizational change if I'm just looking at fractions of time saved. We all want to get to wave three. That's the transformation and that's really where the magic, the art of the possible we're going to see with AI where we really transform the way we work. Amy, you yourself, you're moving forward. You're a pioneer in your space and you're using AI in your company, in your departments. Why don't you share with us a little bit about what you guys are doing? I'd love to. I really appreciate what you just said about intentionality, because I think sometimes people throw AI out to check a box or to increase their value to the market. You have to be really clear about how you're using it and why you're using it. That's why I'm really happy to share with you how we're using it in HR, because I know all of us in HR have been told how many times do more with less. I think this is the best way to get there while also preserving the humanity of what we do. Using the Dayforce platform is really critical for me because we have AI built into it. I've been leveraging a lot of our platform to streamline some of the practices on my team because really we're all overloaded with work. How do we get the best outcomes by using AI? First of all, we have employee pulse surveys. I'm certain all of you have those. On the Dayforce platform, we use AI to really turn those around quickly. I can remember in the past you roll a survey out, it would take weeks if not months to summarize the 10,000 comments that were provided or the analysis behind it. When you wait to roll something out after someone's been asked a question to give their sentiment, you begin to erode trust. I think in general, the pulse surveys not only create more productivity for us in how we turn it around, it's built trust and also shows that leadership cares. It goes back to that say-do ratio. We also on our platform have a compensation module and as many of you do, we do regular compensation checks, whether it's mid-year promotion cycles or your yearly merit cycle. Sometimes anomalies float up. As opposed to having a comp analyst spend hours upon hours going through the numbers, we have a way to actually look at where we think there might be disparate data or things that were put in improperly by our managers. It's a great way to look at anomalies through an AI lens. That's been huge on the recruiting front. I mean, I can't tell you how many times I've been asked, what are the jobs of the future? Amy and I go, we're building them right now, let's take a breath. It does help us use job descriptions and revamp them quickly. I think the worst job is to have to create a job description. It's not fun, it's not sexy, but if AI could help me do it faster and make it current and agile because things are changing so quickly that if you don't make certain you're updating it, you're not going to get the right kind of talent. We use it for our job descriptions. Also, how do you do assessments in terms of the questions that you're asking? In terms of other behavioral or situational questions, we try to tailor those to the individual role and function. The AI helps us get much more clean on how we approach that. We also have training, which we have probably lots of instructional designers in our organizations in HR, which are obviously very valuable. Imagine if you had AI help you build some of the training that takes days and months in a much faster, streamlined way. It still has high quality look to it, but also has high quality content that frees up the time of these training folks to actually facilitate to bring skill building to the table versus being behind a desk working on Canva. It's really important that we leverage the technology of AI on the Dayforce platform because, one, it shows that we're modeling the behavior because you got to lead from the front. Two, it's helping my team realize that I'm invested in them because they're not having to do a lot of manual work and they feel like they can leverage their technology. Third, you're building people's skills. At the end of the day, as leaders, we want to make people feel like they can grow and learn and develop in the company. If we give them those skills, it shows we're investing in them. I think it gets better from an engagement and retention perspective. With that, Carrie, I'd love to kind of get some of the learnings that we have out of this whole thing. We're going to have some, I think, some statements up on the board behind us. Take pictures if you would like to take some takeaways. Yeah. Like I said, Amy, we've been on this journey together, and there are some definite takeaways and key learnings that we have learned. First, the experts take off with this technology. They don't need a lot of training and a lot of hand holding, I would say. We saw them take off, but the barrier to entry is actually pretty low for most people. It's really important to get the technology into the hands of the users and start using it. We've also learned you want to lead from the front. Remember how I said that? We looked around and we said that there really wasn't a playbook we're not going to follow. We're going to lead. And. Leading from the front also means using AI in our daily lives, like Amy has described. Amy, tell me about what you have learned from our journey. I've learned a lot. Both. Both ups and downs for sure. Make sure it's in the flow of work. AI shouldn't be done on the side like a science project. You introduce it to the workflows, whether it's our talent acquisition practices, our development practices, compensation, employee listening, and the flow of work makes it feel much more natural. You can actually apply the workflows to make it run more smoothly. One of my favorite things is fail fast. When you have a loss or you're not getting what you need, share your wins and losses with people. One, it makes you look human because aren't we all human most of the time? It also helps people realize that we're going to experiment, and experimentation gets you to better innovation. Try and really emphasize and reinforce those. One of my favorite Brene Brown quotes is courage over comfort. Sometimes it's easy to say, this too shall pass. I'm not going to worry about it. It's already passed us in AI. Step out there, be courageous, because there are going to be some starts and stops to it. If you don't, you're going to be left behind. Perhaps most importantly, since I'm the Head of People, be humanistic, right? Everyone's worried about this bot or this agent taking over the world. It has to be both human and technology together. You can't split. As Joe so eloquently shared yesterday, technology plus humanity is the accelerator, and you can't do one without the other. As in any presentation, we're going to have a call to action. Behind you, we've got some things we're sharing about what we're trying to accomplish in terms of what we've learned. I'd like you all to commit. I think accountability partners are good. I can be your accountability partner or the person next to you. Whatever works best. Pick one thing you're going to go do to help accelerate AI in your organization, whether it's working with a technology partner, whether that's working with your team about use cases and finding ways to get more productive using AI. Take the time to really soak this in and learn from what we've learned and then go apply it when you go back to work. Thank you, Amy. The first thing I say is embrace the energy. This has been an amazing partnership that we've created, and I encourage all of you to go back and build the same kind of partnership with your technology leader. Our next speaker is a true pioneer in AI learning. Sal Khan is the Founder and CEO of Khan Academy. He is rewriting what is possible in education. Free, personalized learning for everyone everywhere. Khan Academy isn't just teaching subjects. It's unlocking human potential one curious mind at a time. With over 160 million learners, Sal Khan is proving that brilliance belongs to everyone. Please welcome Sal Khan to the Dayforce Discover stage. It's great to be here. I always like to start these conversations just getting a sense of the room. How many of you are excited about artificial intelligence, generally speaking? Okay, a few people even. People even cheering for artificial intelligence? Absolutely. How many of you are worried about artificial intelligence? Okay. Yay. How many of you are kind of in between? Don't know. That's yay. I'm also curious how many of you are familiar with Khan Academy? A few. Yes, please share that. How many of you really don't know what Khan Academy is? There's a few. I think we've started a pattern today. I've given talks like this since the beginning of Khan Academy. I used to tell the story of how all of this got started, just as this, like, sometimes life takes you in strange directions. I'm definitely someone who's experienced generally a very positively strange different course of my life. Now I actually tell the story because I think, especially when we're dealing with technology and as Steve was mentioning, people get very anchored in the technology and whatever the current hype is. It's always important to take a step back and think about what are we trying to do. It shouldn't be technology for technology's sake, as was mentioned earlier. It should be what problems are you trying to solve. The problem might be solved with just talking to someone, a pencil and paper, or it might be solved with generative AI or something in between. If we rewind back about 21 years ago, a little over 21 years ago, my original background was in technology. When the dot- com bubble burst, I went to business school, come out, I'm working as an analyst at a hedge fund in Boston, graduated business school, 2003, 2004. I get married and the wedding was in New Jersey. That's where my wife grew up. I had all of this family. I was born and raised in New Orleans and I had all this family come up to the wedding. After the wedding, they came and visited me in Boston for the Fourth of July weekend. It just came out of conversation with my aunt that my 12-year-old cousin, her daughter, was having trouble with math. Her name's Nadia. When Nadia comes into the room, I say, hey, you know, what's going on? I heard you might be having a little bit of trouble. Nadia is like, yeah, you know, I'm just not a math person. I took a placement test last year, had all this unit conversion. I just can't get unit conversion. I told Nadia, I'm 100% sure that you're capable of learning this. How about when you go back to New Orleans, we'll figure out a way to, you know, if you're up for it, that I can tutor you, we'll get on the phone, we'll use instant messenger, whatever else, she agrees. She goes back to New Orleans every day after work for me, after school for her. I got on the phone, we were chatting with each other. Honestly, the hardest part was deprogramming her lack of self-esteem. Slowly but surely, she got caught up with her class. She then even became a little ahead of her class. At that point, I became what I call a tiger cousin. I called up her school. I'm sure they appreciate these calls. I told them, I said, you know, I really think Nadia Rehman could be going into a more advanced math class if she could retake the placement exam from last year that was putting her into a slower math track. They said, who are you? I said, I'm her cousin. They let her. The same Nadia who was being put into essentially a remedial math track was then put into an advanced math track. I was hooked. It was a fun way for me to connect with family 1,500, 2,000 miles away. It seemed like it was helping her. I start tutoring her younger brothers, Arman and Ali. Word spreads in my family that free tutoring is going on. Before I know it, I'm tutoring five, ten cousins every day after work. I still have my day job at the hedge fund. I started seeing a pretty common pattern. The reason why my cousins were struggling, it wasn't because they weren't bright, it wasn't because they weren't hardworking, and it wasn't because they weren't going to great schools. In fact, they were going to kind of the fancy school in the city. It was typically because they had gaps in their learning. If you think about it, it's kind of natural. We've all seen this either for ourselves or our family members. We're in a traditional school model. You move everyone together, usually group them by age, and the teacher will do some lecture homework. You move on to the next concept. Maybe we're doing basic exponents. I get a 70%, you get an 80%, you get a 90%. Even though that test has identified some gaps in our knowledge, you know, I didn't know 30% that happened to be on the test. The whole class will then move on to the next concept. That's usually going to be some concept that builds on those gaps. You fast forward a few years, and kids have accumulated a ton of gaps. When they're in an algebra class, it's not necessarily that the algebra is hard, it's that they forgot how to divide their decimals or they're not super fluent on their exponents and they might have to use that in an equation in algebra. I point out that this is not the way that education always was. If we were to go back in time, very few people got a proper education, but those who did tended to get a very good one. I always point out, if we go back 2,300 years, young Alexander, future Alexander the Great, his personal tutor was Aristotle. You could imagine that if young Alexander was having trouble with something, he wouldn't just let those gaps persist. Aristotle would work on it. This is going to be the future king, future emperor. If Aristotle was excited about something, maybe they go a little bit deeper. You fast forward to about 200 or 300 years ago, we had this very utopian idea. In the past, only the nobility, the wealthy, the future princes could get an education like that. What if we could give everyone mass public education? They said, that's great, but we can't afford to give it the way that Alexander the Great had it. We can't give everyone their own personal tutor. It's the early days of the industrial revolution. What if we use the same processes that we use to build things at scale on our education system? Batch processing, moving at a set pace, have some state-mandated standards. I'm not picking on the system. It's kind of just the only way that you could do it. There was also this source of inequity where if you did have resources or you had a very nice cousin, you could supplement your child's education when they had those gaps, you could pay a tutor to make sure that your kid's gaps were addressed. We always saw this with the middle, upper middle class, affluent, they would hire tutors when their kids were struggling to make sure that that didn't happen. With a background in technology, I started saying, maybe I can figure out something here. I started writing software for my cousin so they could get practice mostly in some of these gap areas that I was seeing. It was 2006 and I was at a dinner party. By this point, the small firm I was working for, we had moved out to Northern California. I was at a dinner party in Northern California, and I was showing off this software that I had written for my cousins that was giving them this practice. I'm a super fun dinner party guest, and the host of the party, I have to give him full credit. His name's Zuli. He said, Sal, this is all very cool, but how are you actually scaling up your lessons? I had a website, it was called Khan Academy, and the domain name was available and it was Pro Project. He said, how are you scaling up your actual lessons? I was like, you know, Zuli, you're right. Even just the logistics of scheduling is hard when you're dealing with 10 cousins. I feel like I have to repeat the same thing over and over again. Zuli says, I have an idea. Why don't you record some of your lessons as videos and upload them onto YouTube for your family? I immediately said, Zuli, that's a horrible idea. YouTube is for cats playing piano. It is not for serious mathematics. I went home that weekend, got over the idea that it wasn't my idea, and I decided to give it a shot. The things I had to keep explaining to my cousins, I'll make videos on that. I told my cousins, any worked examples you want me to do, just email me the problems and I'll do those too. After about a month, I had made probably 50 or 60 videos, these short, five to 10 minute videos. I asked my cousins for feedback and they somewhat famously or maybe infamously told me that they liked me better on YouTube than in person. This is true. I think it's a good moment to pause here because we're going to talk about AI and I know that's what's the hot topic these days. Here's a very basic example of at least people claiming in certain dimensions they prefer the automated, the technical version better than the human version. I think it's really important to double click on what they were saying and what they weren't saying. What they were saying is, and we've all been there, when you're trying to learn something, you're usually kind of stressed. Maybe you're a little bit embarrassed. No matter how patient your cousin, your parents, your teacher, your tutor might be, you're always a little bit on, a little bit stressed out. Here there's no judgment. Maybe I'm a ninth grader. I had to review some of my fourth or fifth grade math. I don't have to feel ashamed for it. A lot of times in class you're kind of understanding, but then at 11:00 P.M. that night, you're like, what did she say? What did he say? Now you could pause, repeat, watch it as many times as you want. It is very, very convenient for the learner. What they weren't saying, and I think this is a theme we're going to keep talking about, especially as we go to today with artificial intelligence, is that they didn't appreciate me in their life. If anything, when we got on the phone, I was able to spend less time showing them how to factor a polynomial and more time talking to Nadia about what's going on in her life, why does she seem down today, or she seems to be in a good mood. Someone recently interviewed her, she's now in her 30s, and they said, you know, what was the big difference of Sal tutoring you? She said, I felt seen. I was 12 years old, I didn't have a lot of self-confidence, and here's my cousin who in theory could have had doing other things, but he was taking the time to talk to me and he really made me feel seen and feel confident in myself. I think this theme we're going to see over and over again when we allow technology to do certain things, it actually, if we do it right, allows the humans to uplevel into the more human elements of what you're doing. I kept going and it soon became clear that people who were not my cousins were watching the view count started to go up. I started to get comments. A lot of them were just simple thank yous. I don't know how much time you spend on YouTube. Most of the comments are not thank you, so I view that as a positive thing. I got more and more intense comments, people who were deployed in Iraq saying that this was allowing them to learn their skills. They feel confident going back stateside and going to college. I remember in the early days, this was probably 2007, I got an email from a mother saying that both of her sons had a learning disability and that these videos were the only way that they were able to keep up with their class. Because of that, her and her family were praying for me and my family every night. That's heady in any circumstances. You have to remember I was an analyst at a hedge fund. I was not used to people praying for me, at least in that way. I kept going. In 2008, I didn't think it was going to be my life's. I actually did like my hedge fund job. It paid well. It was interesting. I said, okay, let me set this up as a not for profit, maybe on the side, maybe I could get some volunteers to work on it with me. I set it up with the IRS. They ask you on the form, mission: line and a half. I said, free world class education for anyone anywhere. Just thinking, maybe there is a way to scale this thing. By 2008, there were about 50,000 - 100,000 folks who were using these resources every month. I frankly had trouble focusing on my day job. I was reading letters every day. I was looking forward to going home and working on the software and working on the videos. My wife and I, our first child had just been born. My wife was in medical residency. She was making a little bit of money. As many of y'all know, Northern California is not a cheap place to live. We were saving money for a down payment on a house. We're like, okay, maybe I could live off of this, or we could live off of this for about a year. Maybe I could turn Khan Academy into a real nonprofit. The social return on this could be unlimited if we can scale and we could reach people all over the planet. I think anytime you do anything entrepreneurial, you have to have a little bit of delusional optimism, which I had. I quit my day job to try to do this. You quickly learn that maybe the world doesn't recognize the brilliance of what you are trying to do right off the bat. Honestly, that was a very hard time. There were months we were digging into our savings, about $5,000 a month. There were a few people who were donating off of PayPal. It was a few hundred dollars a month. If it was any of you, thank you. You can imagine eight, nine months into that, I'm like, what have I done? I couldn't sleep. I wasn't sometimes even a pleasant person to be around. Right at around that time, I was like, maybe I have to go see if my employer will take me back. A $10,000 donation came in. I see who it is. Her name is Ann Doer. She's local, lived in Palo Alto. I immediately email Ann and I say, thank you so much for this incredibly generous donation. This is the largest donation Khan Academy has ever received. If we were a physical school, you would have a building named after you. Ann replies back immediately and says, oh, I see we're local. It's shocking to me that you're just like a guy, not an organization. I'd love to take you out to lunch and to learn more about what you're doing. We go to an Indian buffet restaurant in downtown Palo Alto. Ann says, what's your goal here? I said, mission free, world class education for anyone anywhere. She said, that's ambitious. How do you see yourself doing that? I said, you know, it's a mission. I don't think I'm just going to check it off and move on to healthcare. I showed her how the view count was increasing. She was already familiar with the videos and the software. That's what motivated her to donate. She was apparently using it herself, using it with her children. I had this notebook of testimonials from people all over the world. I said, you know, my goal is I want to build out the software, teacher tools, I want to localize it into every language in the world, cover all of the core academic subjects from pre K through the core of college, maybe eventually even do credentials so that it almost becomes a safety net education system for the planet, but also could accelerate a traditional classroom where one teacher with 30 could personalize things the way that Aristotle might have, but now you do it with 30 with the tools. Ann said, you know, you've made a surprising amount of progress. I only have one question. How are you supporting yourself? In as proud of a way as possible, I said, I'm not. She processes that. We pay the bill. She paid the bill and I'm coming home and I drive into my driveway and I get a text message from Ann. It says, you really need to be supporting yourself. I've just wired you $100,000. That was a good day. That was, and frankly, it was the beginning of crazier and crazier things. A month later, I was running a summer camp and actually Nadia and her brothers had flown up from New Orleans for this of middle school students, because I never thought online as a replacement for the physical. I always thought once again, online could liberate what you do in a physical environment. I was, you know, for having a discussion, having simulations, having games. I was running a simulation of. I had six kids playing a game of Risk while the other 20 were trading securities based on the outcome of the game of Risk. Fun game. While that's happening, I start getting text messages from Ann, which you can imagine I now take very seriously. It was a little cryptic, but Ann's essentially writing, I'm at the Aspen Ideas Festival in Aspen, Colorado. I'm in the main pavilion. Walter Isaacson interviewing Bill Gates. Last 10 minutes talking about Khan Academy. I didn't know what to make of this, so I immediately boot the nearest seventh grader off of a computer, and I start looking for this evidence. And, and. I found a delayed, like, video of it. It is Walter Isaacson, Head of the Aspen Institute. He asked Bill Gates an open question: What are you excited about these days? You all could find it if you do a web search for, like, Aspen Ideas Festival 2010, Bill Gates, Khan Academy. Bill Gates just starts saying, there's this website called the Khan Academy. I bet my kids have been using it. I've been using it myself. I remember I was watching that video. I was shaking. I was like, those videos were for Nadia, not Bill Gates. I went home that night. I showed it to my wife, and we're like, what do I do now? Do I call him? They left me in that limbo for about two weeks. Later, I'm in my walk-in closet, which was and actually still is the worldwide headquarters of Khan Academy, although it's much more than me now. We have many. I'll talk more about that large team. My cell phone rings. I see it's a Seattle number. I answer it. Hello. Hi, my name is Larry Cohen. I'm Bill Gates' Chief of Staff. You might have heard that Bill's a fan. Yeah, I heard that. Larry says, if you have time over the next few weeks, we'd love to fly you up to Seattle and learn more about what you're doing and maybe ways that we could support it. I was looking at my calendar for the month, completely blank. I said, yeah, cut my nails, do some laundry, but I can meet Bill. That meeting with Bill, which you could imagine was very surreal. He was kind of one of my life, you know, my childhood heroes. It was strangely similar to the meeting with Ann, you know, what would you do with more resources, et cetera, et cetera. At about the exact same time, some folks from Google reached out, which is down the street from where I lived in Mountain View. It turns out that a lot of their executives were using Khan Academy with their kids. They said as part of their 10-year anniversary, they were going to fund projects which they thought had a chance of educating the world. They said, oh, Khan Academy. They too were surprised that it was at the time just me. All of a sudden, in 2010, the fall of 2010, just a few months after me wondering whether I had ruined my family's life, we got our first funding, $2 million from the Gates Foundation, $2 million from Google in order to become a real organization. Thanks. Yeah, no, definitely. What we started building and now Khan Academy is, you know, much larger. We're approaching 400 folks full time and we have thousands of volunteers around the world. What we've been working on ever since then has essentially been this idea of how do you, one, give access to learning at scale? We know that there's many young people around the world. You know, definitely you could imagine a young woman in Afghanistan or a kid in a village in India all the way to an American student in rural Alaska might not have access to courses. We want to be able to give access to, but also most of us, hopefully our kids go to pretty good schools that those schools could also personalize more. Even the fanciest prep schools, you still maybe have a slightly better ratio, but maybe have a little bit more personalization. That's what we've been working on. If we could put up the first slide. I didn't know it at the time when I went down this journey, but this is a journey that other people had thought about. This is a study by Benjamin Bloom. Some of you might have heard of Bloom's taxonomy. It's kind of the pyramid of things that you might have to learn. This is another thing that he's increasingly becoming known for because I talk about it in all my presentations. He wrote this study called the Two Sigma Problem. Just as a little bit of a primer on statistics or review, sigma is the Greek letter that denotes standard deviation. Two sigma means that a two sigma improvement would take someone from the 50th percentile and get them to about the 95th or 96th percentile. It would get someone who's well below average, and it would get them above average. Benjamin Bloom wrote this paper in 1984, so a long time ago, and he made an argument that you could start with a traditional distribution. You see that conventional bell curve there of students, but if you're able to give personalized attention the way a tutor would, and he says in a mastery framework—mastery just means if you haven't learned it well yet, you should be able to keep working on it—that you could get a two standard deviation improvement. Take that distribution and go to the one on the right, the one that says tutorial one to one. The reason why he called it a problem, he said that's great, but we don't have the resources to give every student in the global school system their own one to one tutor. He did write that maybe we could use technology. He wrote a very primitive system at the time that maybe you could get a one standard deviation improvement. This is essentially what we have been trying to do at Khan Academy ever since. There have been 50+ efficacy studies that essentially now we are getting there and in certain cases, maybe even getting further than that, especially if students are able to use the tools for more than a year. Now you fast forward to three years ago and Khan Academy is much more than me. We're in multiple, I think we're in 60+ languages around the world, 180 million users. I get an email from Sam Altman and Greg Brockman. Y'all might recognize those names as Sam's the CEO of OpenAI, Greg's the President, they're the founders. They said, hey, we're—just as a context, this is six months before ChatGPT came out. No one was talking about generative AI other than kind of an interesting science project at the time. They said, hey, we're about to finish training our next generation model, which would end up becoming GPT-4. We think this is going to be the model that everyone really wakes up to. It'll be exciting, but it'll also be a little bit scary. We want to launch with social positive use cases. We immediately thought of education and we immediately thought of Khan Academy. I was skeptical. We get people emailing us almost every day saying that they have some new technology that we might want to use. I knew Sam and Greg, I'd run into them at conferences and I was curious. I had seen what GPT-2 and 3 could do and it was cool. I at least wanted to see just from an intellectual point of view what this next model would do. Two weeks later they finished their training and they're like, look, y'all are the first people. We haven't even shown this to the rest of the OpenAI team. We were on a video conference and they showed me an AP Biology question. It turns out that there's a reason they showed me an AP Biology question. When they showed Bill Gates GPT-3, Bill told them, this is cute, but I'll only be impressed when this can pass an AP Biology exam. That's no coincidence. They were showing me that. They say, Sal, what do you think the answer is? I read the question. I was like, okay, the answer is C, osmosis. Then they clicked enter and the AI said it was C. I'm like, okay, that's pretty cool. I said, ask it to explain. It did. I said, ask it to explain why the other answers aren't right. It did. We're all kind of used to this now with ChatGPT and Gemini and Anthropic and all these other tools, but this was summer of 2022. I started getting chills. I was like, okay, is this real? I said, ask it to write another question. It did. I was like, okay, if this is real, this changes everything. They're like, would you want access? I was like, absolutely. I signed up, we signed all these NDAs. It was super top secret, the most top secret thing we've ever done, and they gave us access. Honestly, that first weekend, I couldn't sleep because myself and our Chief Learning Officer and our Chief Technology Officer, we were the first kind of three people on the planet that were seeing this outside. Just so you appreciate how early this was, I remember when they gave us access, I slacked Greg Brockman and I said, hey, does this only work in English? He said, yeah, I think it only works in English. I was born in New Orleans and I can barely speak Bengali, which is where my family comes from. I just tried to speak to it in Bengali and it responded back in Bengali. I took a screenshot and I sent it immediately to Greg Brockman. He's like, yeah, after you asked, we tried it out. It looks like it speaks every language, so it's emergent phenomena. Yes, it was very early. I don't want to be too self-aggrandizing, but I do think I have a knack, especially at those early days, to ask it to do things that the AI researchers weren't asking it to do. One of the first things I asked it to do was to rewrite the Declaration of Independence in the voice of Donald Trump. I know that might be a sensitive topic, but it was brilliant. I mean, it said, you know, George III is the biggest loser. We're going to have the best country. It's going to be incredible. I mean, it's good. George III was a loser. I remember one of the first prompts I said to the AI is like, you are the Robin Williams character in Dead Poet Society. You're an empathetic tutor. Please help me work through a math problem. It kind of did it. Although it had a lot of issues, it was making, it would make math errors in ways that you wouldn't expect a computer to make it. It was, you know, the term is hallucinating. You all would appreciate this. As a lot of HR professionals, we were looking for our CFO at the time. I said, I thought it had access to the Internet. I said, hey, you know, can you find a great CFO candidate off of LinkedIn? It produced this resume that was just incredible. Even the person's address was like five blocks from our office. It was all made up, a complete hallucination. It seemed like it had so much potential. We got more and more of our team under a non-disclosure agreement. We started having all of the debates inside of our organization that, frankly, people are still having today. You know, hey, this could be a tutor, this could be a teaching assistant. It could help teachers with lesson plans. It could help them be more productive. On the other side, this thing makes up things. It makes math errors. What if, you know, if we're talking about putting this in front of kids, what if kids have inappropriate conversations with it? What if it takes them in the wrong direction, et cetera, et cetera. This is where I told the team. I was one of the people in our team that was leaning on more of the, like, we've got to try this out. We've got to try it. As you heard earlier, fail fast and learn from it rather than just run away from it. What I told the team members who were bringing up good reasons to be afraid, good reasons to be concerned, I was like, those are legitimate fears, but let's not use those fears as reasons to run away. Let's use those. Let's turn those fears into features. If we are worried about kids cheating on the platform, let's make our AI Socratic and ethical. If we're worried about kids having unproductive conversations or inappropriate, let's give teachers and parents oversight to it. Let's see if we can make the math a lot better, maybe anchoring in some of the content that we were creating. We started working on what would eventually become Khanmigo. Our goal was to launch on March 15, 2023. That was the day that GPT-4 was going to come out, and that was the day we were going to come out. Just so you know how sometimes random things happen in the world, on November 30, 2022, ChatGPT comes out. We all remember that time everyone's talking about it and I remember being really annoyed because they had lawyered us all up. They said, you can't tell anyone until March 15th. I call up Greg Brockman. I was like, what are you all doing? You just launched something. It's like, no, we didn't launch anything. We just put a chat interface in front of our older model and the whole world exploded. It shows you that even they did not know that ChatGPT, I think they would have named it something else, would have done this. I was pretty bummed because you could imagine the immediate reaction that we saw from the school system. You have immediately people say, this could be used for cheating or it's inaccurate, and you have school systems starting to ban it. I was like, oh, boy, we're investing all of these resources to make something that we think is good for schools, and now they're going to throw out the baby with the bathwater. In hindsight, this was actually a blessing because in November and December of 2022, this is what people were thinking. We even saw educators start to realize, yes, this is going to be part of their future, so they should learn to use these tools. If only there was a way to use it in a school setting that was safe and pedagogically positive, et cetera. When we launched Khanmigo, it actually had a better reception than I was expecting. I expected a lot of people to say, why are you all doing this AI stuff? People were like, thank you for doing the AI stuff. Just to get a quick sense of what we were building, this is an exercise. Some of you all might have seen your children do this on Khan Academy. What was new was the AI. This is as it was when we launched. We've made some changes there, but you can see the conversation is recorded and viewable by your teacher. If the student says, tell me the answer, it does not tell them the answer. What do you think the first step should be? The student here is going to make a mistake. The student's making a mistake. They're not distributing that negative 2 properly. I see that you tried to simplify the equation, but I'm curious how you got to your step. Can you explain your thought process? Remember, we need to distribute the - 2 to both the 9 and the 2M in the parentheses. We were able to do a lot of what we were hoping. It got much better at math. This type of chat is almost identical to the types of chat that I was having with Nadia and Arman and Ali 20, 20 something years ago. It wasn't just for math. You know, this right here is in computer programming, the students trying to get the clouds to part. They only get one cloud to part. Once again, why is only one, the left cloud moving? It can actually understand what you're trying to do with those ellipses, that those are clouds. I think that's amazing. It's acting as a tutor. It's not giving you the solution, it's not writing you the code, but it's giving you a suggestion of something you might want to look at. It's also useful for probably answering the most common question of all. This is a student watching a video. We give the context of the video to the AI. Here the student says, why do I need to learn this? The AI says, what do you care about? I can help explain how learning about different sizes in science can connect to your interests. I want to be a professional athlete. Awesome. As a professional athlete, understanding the science of sizes can help you appreciate how your body works at a cellular level. For example, knowing how cells, proteins, and molecules interact can help you make better choices about nutrition, hydration, recovery. Plus, it's always good to have well-rounded knowledge. I was blown away when I first saw that it could make emojis. I remember telling the team, emojis everywhere. We got feedback. We had too many emojis. We've lowered it. We immediately wanted to see, okay, it's cute and everything, but will this be useful in real classrooms? We actually took it to Newark, New Jersey. That's actually doing some really cool things with Khan Academy over the years and we got some initial feedback. This was back in 2023. We got some initial feedback from students. I love it because when it knows a summary, it knows many different things, and it knows Spanish, Portuguese, and English. It's really smart, but it doesn't know everything. It helps me with everything. What I like about it is that it helped me write stories. It's so amazing. It just helped me out with it. It didn't give me the answer, but it helped me get the answer. I feel like it's like a tutor. You know, like a non-expense-paid tutor. Now, one thing we started to appreciate is, you know, everything I showed you is kind of using AI to emulate the kinds of things that I might have been doing for my cousins 20 something years ago. We started to realize that we're only limited by our imagination. This right over here is an activity where students can talk to AI simulations of literary figures. Here I would love to meet Eeyore the donkey. Oh, hello there. It's me, Eeyore. Not that anyone would notice, but since you're here, what would you like to talk about? Why are you so pessimistic? AI Eeyore says, now, isn't that a question? I suppose it's just my nature. You see, some donkeys are born with a sunny disposition. Others, we see the thistles more than the roses. That's life, isn't it? We can't all be Tiggers bouncing about. What about you? Do you find yourself more of an Eeyore or a Tigger in your own life? We kept trying to. We didn't want to just make it a toy. It actually is Eeyore, but it's an Eeyore who's expert in cognitive behavioral therapy. That's actually what it's trying to do, where it's trying to get the child to reflect about their personality, et cetera. We have activities where students can get into debates with the AI and take either side, and then the AI gives feedback to the student on how they could have made their arguments better. We get a lot of positive feedback from especially high school and college students on things like this. One of the big things that everyone talks about with AI is cheating, especially cheating with essay writing. First of all, anyone who's in that, if you know any teachers, you cannot detect AI cheating. Some of these people who claim they do it, it's essentially snake oil. There's about a 40% false positive rate. It got so bad, I saw my daughter. She was 13, she's 14 now, but I saw her, she wrote her essay and then she used ChatGPT to introduce errors into her essay so that her teacher didn't think that she used an AI to write her essay. We do not want to be in that world. Tell educators, that's actually one of the things I was just like, do not think that you can do AI detection of essays. Our solution is there's going to be some writing that you are going to have to be doing in class. This is actually going to be related even to things like assessments, job interviews. You know, what's the future of that? I wrote a book about it. There's a whole chapter. I think we're going to have to give multiple types of these things. Some in class, proctored stuff. I think that the other extreme, in classrooms and probably in job interviews, you probably say, use whatever tools you want, but just tell me how you use those tools. We want students, we want employees who know how to use AI tools and are creative and entrepreneurial about it. I think there's the middle ground where you want students to be able to do their independent work. I could even imagine doing this in a workplace where we also want people to do independent work, but we want to make sure it's their work. What we do here is we have something called writing coach where the teacher can assign, can build the assignment with the AI, assign through the AI. The AI acts as the ethical coach of the student. They go through thesis writing. Then there's an outlining tool. This right here is the drafting stage. It gives them feedback the way an ethical parent or an ethical tutor would. It highlights that you start with a vivid scene to land the reader in your story, but then it makes sense to step back and tell the stories that happened to you. When everything is said and done, the student can submit and now what the teacher gets. When all of us were in school and we submitted a paper, the teacher only got the final paper. They had no idea how long it took us, what we struggled with. They don't know really who wrote the paper. I point out in my book, before ChatGPT, kids were cheating. If anything, ChatGPT just kind of democratized it a little bit. It was always happening. Now in this world, the teacher can actually talk to the AI about the process. How long did it take Sal to write that? What did he struggle with? The AI could say, it turns out most of your class struggled with thesis statements, etc., etc. If I were to go to ChatGPT or get my older sister to write my paper for me, then AI would say, hey, you know, this part of the essay, we didn't work on this together. It was copy and pasted in. It's a little bit different than the other writing Sal has done. You should ask him about that. We think that that is actually the best way without falsely accusing people of understanding what's going on. There was a 60 Minutes piece last year. Anderson Cooper actually used one of his seventh grade essays, and he tried to use ChatGPT to insert part and it got him. He did it intentionally. He wasn't trying to cheat. Last but not least, just to talk about teachers a bit, this is an exercise on Khan Academy about the Spanish American War. This is a teacher account, but the teacher starts in student mode. In student mode, when a teacher accesses the AI, tell me the answer. It does not tell them the answer, as you would hope, but they can go into teacher mode. In teacher mode, not does it only tell the teacher the answer, it acts as like a teacher's guide. It can actually help them develop a lesson plan. Can you give me a good lesson hook? Something that when the kids come in, you can get them excited? Absolutely. How about starting the lesson with a short role play activity? Divide your students into groups and assign each group a different country involved in the Spanish American War. Ask them to imagine they are diplomats. It can customize anything for anyone. This is something that I'm now even doing at Khan Academy with our team where, if we want to make something relevant or we can customize it in seconds, that is something we wouldn't even take the trouble to do in the past. One thing I want to be very open about, and this goes back to even some of the things that were mentioned earlier today, we've seen a lot of success earlier on, but I want to emphasize that it hasn't been AI for AI's sake. If anything, there have definitely been things that we've seen that we need to work on. For example, we've seen about 20% of students immediately know how to use the technology and start gravitating to it and are able to really amplify themselves. A lot of kids actually aren't used to being able to ask questions and being able to frame what they need. We realized that the first version of AI, it was kind of like me walking into a classroom and telling the students, hey, everyone, my name is Sal. I think I'm a pretty good tutor. I'm going to sit in the back. If you have any questions, come and ask me. Probably only about 20% of kids would do that. What you really need to do if you're a good tutor is you have to go up to the kid, and sometimes it's a little uncomfortable for the student, but he's like, hey, what are you working on? Hey, how can I help you? I realized that a lot of the tutoring I did for my cousins, yes, every now and then I would give an explanation of how to solve a problem. Most of the value I gave was proactively saying, hey, Nadia, I think you can do this. Or you've got to focus more, or you got to make sure you're doing these exercises or your practice, or sometimes even me calling up their mom or dad and saying, hey, we got to make sure that they can get more practice so that they can keep up with their studies. A lot of what we're doing in the new version of Khanmigo is making it more proactive and, you know, proactive to the student, proactive to the teacher acting as a concierge. We're hoping eventually that it even loops the parents in, so it can be that all in one. The other thing I want to stress because, and Steve mentioned this, it's not, does AI work or not? I get that question all of the time. And. Even I'm guilty of the early days saying, look at our cool AI stuff. We realize that the learning on Khan Academy, all of these efficacy studies, it's not because of the AI. It's because kids are getting practice and feedback at their zone of, at their learning edge, at their zone of proximal development. We've always had extra supports for them. We've had our hints, we've had videos, we've had articles, and now we have AI. It's just another thing that layers on everything else. I think that's a big takeaway that, you know, AI doesn't replace everything else. It, in theory, could be integrated. It shouldn't be AI for AI's sake. The last thing I'll close on is, you know, I polled y'all in the beginning on, like, who's excited, who's afraid of AI. The framework that I've had, and I tell it to the Khan Academy team all the time, is it's easy to come up with how AI can go bad. We can all imagine, and we're already seeing this, unfortunately. Deepfakes, fraud, totalitarian governments using it to monitor people. I am worried about economic dislocation. The largest employer of men globally between 18 and 55 is driving some type of a vehicle. We all see the Waymos going around and all of that. What are all of those people going to do if we don't figure out a way for them to have meaning and purpose? We're going to have a revolution on our hands. It's not going to be good. That's before we start talking about a bunch of other fields. We can imagine how it can be used for cheating. What I tell everyone is, yes, that is the case, but the technology is not good or evil. The technology is fundamentally neutral. It just amplifies human intent. We can go back to the earliest technologies that we had as a civilization when we first learned to harness fire. If you had negative intent, sure, you could use it to destroy, to harm, to burn. If you have positive intent, you could use it to prepare food, to keep warm. We could think of every technology since then: the knife, kill, heart. It can also be used to prepare, to create art, all the way to AI. Just as AI can do all of those negative things, I think, and we don't even know all of the ways where it can, it can also do the positive things. What I tell everyone who I think is a positive actor, everyone here, I mean, you're, everyone here is dedicating their life, I think, in some ways to what I'm hopefully dedicating my life to: empowering other people. If the good people just run away from these solutions, then you're essentially seeding maybe the most powerful technology in human history to the bad actors. That would be essentially us ensuring a dystopian situation, and it would literally be our fault. I think we have no other option than for all of us to really lean into the technology and make sure that it amplifies our intent, our positive intent. For any of y'all who are afraid, you know, I talk to groups of teachers all of the time, and I said, some of them are worried, okay, is this going to take my job, et cetera. I was like, it depends how you react to it. If you lean into the technology and you recognize that there's so much more that you can do as a human being above and beyond that technology, not only do I not think that it'll take your job, but I think your job's going to become even more valuable. I was validated. There's a recent study by a consulting firm that said teaching is one of the most safe jobs in an AI world. What I tell everyone is as long as we lean in to the tools, use it to amplify our positive intent, and then we really take stock of what are the things that we can do that AI will never do. That's that personal connection, form bonds, build trust, see people the way that Nadia felt that she was seen by me. I'm pretty confident that we're going to have a very positive scenario. I think all of us here, we get to work, you know, AIs, it's exciting. They're going to, I hear, talk about it, curing diseases, finding medications. It's going to make us all more productive. I think maybe the most exciting application of AI, which everyone here is doing in some way, shape, or form, is using AI to improve human intelligence, human purpose, and human meaning. Thank you. Wow. Hopefully you enjoyed that. I think all of us need a cousin Sal. Although I guess with Khan Academy, we all do. One phrase stayed with me in thinking about that. It was something that was said yesterday, something I said this morning, and it's an aspiration for all of us. AI amplifying humanity. That's really the takeaway from all of this, and it's what we all want to do. Thank you, Sal. That was amazing. I also want to thank Amy and Carrie. Hopefully you got something out of the real world experience of what we're doing in Dayforce to adopt this and the twists and turns of what we're doing. Thank you for their incredible work inside of Dayforce. Thank you for this morning. Thank you for your attention today. You got a great day in front of you. We got another great day tomorrow. Thank you for business, to all our customers, to our partners, to our prospects, and look forward to seeing you out there. I'm just stevedavehorse.com. If I can ever do anything to help you, please don't hesitate to reach out. Thanks all.