0:02So happy you're here to join us to chat about a very important issue. Uh my name is Dan Meyer. I'll just say a brief word about my biases where I'm coming from my priors. I was a public high school math teacher which gave me a lot of appreciation for the school as a social unit of learning. I also work now for Amplify making technology including our first AI feature um a discussion tool supporting uh teachers and whole class discussion. So if you love AI, if you hate AI, I am bound to disappoint you.
0:29But those that's the frame uh that I'll be bringing questions to the panel with.
0:33Um I wanted to share with you who these folks are. I've only just met them today, but I spent um a lot of time aka the time in between the first snack service and the second snack service on the plane researching these folks and getting myself excited about the various resources that they'll bring to this issue. I might be wrong in what I say or certainly incomplete. They're welcome to correct the record once it goes to them for their part here. I'm going down the line here. Um, Caitlyn Mint is a a sophomore um at Stanford studying junior now.
1:01Junior, I'm so sorry. Uh, this is great though. This is the process. [laughter] The process works. Um, a junior studying math after my own heart. Um, we have we're very lucky to have Kate here. Kate has um a couple distinctions that are relevant here. I think one is writing about emerging school models from a a very widely read Substack Austin scholar and also having been um a student at Alpha School um a school that is very AI forward and certainly qualifies as an emerging model. So we'll all benefit I hope from that kind of experience. Um
1:31down the line we have Julie uh Julie Squire uh who is a partner at Bellweather and um thinks a lot about the bundle of resources and people and buildings known as a school and how to creatively reconfigure those for the benefit of kids. um and teachers and people also you uh you worked um at the New Jersey DOE um on something that in my head is like technologyish um which I'm really hoping we can draw on uh during the race to the top years.
1:56We need a lot of that uh experience here on this panel. Um and then Stephen, you're the director of of comms at City Fund. And what's exciting about what I understand about your work is that you are supporting school leaders across city funds uh portfolio of schools in in communicating about issues of this sort.
2:14Um among which right now include enthusiasm and skepticism about AI and a technology backlash that's uh in full full swing. So desperately needing a lot of comm support. I'm sure that's our crew. Would you um give them a warm welcome WITH ME HERE? [applause] UM, I'M GOING TO PASS IT DOWN the line for some prepared remarks. I'll have a few questions of my own and then kick it to you folks for some questions, too.
2:40Um, so first up, uh, Kate, would you please?
2:43Yes, of course. Um, is this Can y'all hear me?
2:47Okay, amazing. So, hi everyone. My name is Kate and I spent eight years as a student at Alpha School. I am now a junior majoring ma majoring in math at Stanford and am applying to pursue a co-term master's degree in education. Over the past six months, I've also returned to Alpha as an intern under Dr.
3:06Tasha Arnold, who is the head of Alpha schools. Returning to Alpha has led me to think about a question that America has, surprisingly enough, faced before.
3:16When technology transforms the world that children will inherit, how should schools change to prepare them? A century ago, tractors rolled onto American farms and began doing the work that had fed rural families for generations. In 1910, one in three American workers farmed. Most of their children would not. The response came town by town. Communities voted to tax themselves, to build public high schools, and write curricula for life
3:45beyond the farm. When they started, one in 10 18-year-olds held a high school diploma. Within a generation, more than half did. Today, we face a similar turning point. Artificial intelligence is entering every workplace, transforming what people do and what it takes to thrive. Yet, our schools still prepare children for the world that AI is replacing.
4:10America needs a new education revolution. AI makes it both necessary and also possible. For a hundred years, it took the whole school day to teach children what they needed to know. But there and there was little time left for the rest of growing up. But now there is. Used well, of course, AI can remake the school day. Every child can master academics more thoroughly at their own pace and in a fraction of the time. At
4:38Alpha, cuttingedge AI personalized learning technology enables students to learn core academics in only 2 hours per day. Then we give children their time back to do fun things like learn to build a go-kart with their friends, write and perform their own play, or plant a garden that helps feed their neighbors. They can find what they love, take on challenges that require courage and independence, and know the joy of contributing to their community.
5:07Teachers gain time back, too. They devote more of their day to each individual child, knowing them, earning their trust, challenging them, and refusing to give up on them. And we are not alone in this. There is an international movement toward expanding the purpose of education. From narrowly from advancing narrowly defined academic outcomes to promoting students academic learning in concert with student mental and physical health, happiness, sense of meaning, relationships, and
5:39a movement toward education for flourishing. The shifting educational landscape, including the onset of AI technologies and unprecedented youth mental health challenges globally, requires us to reinvision how education might nurture students holistic well-being and empower them to make contributions to this changing world. But the most transformative educational innovations emerge with both bold ideas and evidence to prove them. And that's why I'm really excited to be here at Harvard. We have Alpha has just launched a collaboration
6:12with the human human flourishing program at Harvard and we'll be conducting a longitudinal study exploring the impact of educational practices and approaches at alpha schools on students life skills and flourishing. The study will give us insight into our strengths and our areas for growth, empowering us to use independent evidence to develop our approach. Alpha is an innovative model, but we are looking to ground it in evidence so that it so that what we
6:40build is not only bold, but true. We should build schools that help children discover what is worth doing and develop the knowledge, judgment, and confidence to do it. A century ago, America invested in a new kind of school to prepare children for a world beyond the farm. Now we must invest again not in more of the same but in rebuilding school for world beyond for a world beyond routine work. That means letting go of the school that you remember. Farm
7:09families had to do the exact same thing when they sent their children to a high school that almost none of them had attended. The goal is not merely to prepare children for jobs. It is to prepare them for lives of purpose, contribution, and extraordinary possibility. We will raise a generation more capable than we ever were, ready to solve the problems that we couldn't. Let's make this the best time in history to be a 5-year-old.
7:34[applause] All right. Um I um I took the prompt in the title for this session somewhat literally. Are emerging AI models rhetoric or revolutionary?
7:52And as I was thinking about this, I I want to start by just positing that for something to be revolutionary, I think three things have to be true. One, um I think we need to see discontinuous improvement, a notable change in the existing trend. Um I think second, we need to see a change in the underlying rules. Not just efficiency within those rules, not just working outside of those rules, but changes to the rules themselves.
8:22Um and then third I think we need to see scale. So change that reaches beyond isolated pilots or niche providers. So those is sort of like the frame for how I think about what it means for something to be revolutionary. With that thank [laughter] you. Glad it resonates with at least one person in the room.
8:45Um um I think we can apply those criteria then to how AI is being used in schools within the system and then to models that are emerging outside of the system.
8:57Um so my colleagues and I at Belleather have been doing some research on what AI use cases look like within the system and there are five that I want to elevate here. Um and then I'll talk a little bit about how we see AI use cases emerging outside of the system and sort of AI first models. Um but so so five use cases. The first is personalized learning. Um so AI powered tools that can support personalized learning with adaptive learning platforms that adjust difficulty and content based on
9:26real-time student data. Um the second use case would be educator sustainability. Um, so educators can use AI for routine tasks like writing newsletters and time-consuming tasks like um grading open-ended test um responses and spend more of their time in the humanto human relationships that are so essential for student well-being and motivation.
9:50The third use case I would say is assessment. Um, a lot of assessments are restricted to what is easily measurable, but with AI, we have the opportunity to assess different types of skills that are demonstrated in different ways.
10:03Fourth, information and analysis. Um, between student assessments, learning information, learning management systems, student information systems, the data collected by edtech tools, there's a lot of data about students that schools and systems have, but it's often very messy and fragmented. and AI can help connect and stitch those data sources together to drive insights that can inform decision-making at the classroom, school, and system level.
10:28And then the last use case I would say is administrative and operational efficiency. So AI can help develop master schedules. I know every school leader's favorite task is developing the master schedule, right? The bus routes. Um it can help draft reports and write grant proposals. So there's definitely an element of that operational efficiency at play.
10:47So if I think about all of those use cases in the context of discontinuous improvement um and um changing underlying rules and then also the the question of scale and I put it within the school system that exists today. I think there are a number of challenges that are preventing AI technology from being revolutionary as it exists now. So one is all of the challenges that reforms of the past many decades have faced for many decades. Um
11:17challenges in capacity. We can ask people to do things. We can give them the tools to do things. Doing those things well and supporting them to do do those things well is something we've always struggled with. Um making sure that we are um looking at some of the the grammar of schooling and sort of challenging some of those norms as they exist today. um I think is another piece. So there are lots of things that have prevented transformative change in the past that are going to continue to
11:46be barriers to change even in the world of AI. I would layer on top of that some very specific challenges for AI. One is AI literacy for students and educators.
11:57Um another is having the policies in place that establish the guard rails for what AI use should look like. What is safe? Where is human judgment essential?
12:06What are high risk cases and low risk cases? and how do we support people to discern between the two? Um, discerning highquality tools. I have a pet interest in procurement policy, which by the way roots to my time at the New Jersey Department of Education, where procurement was the bane of my entire existence in the edte space. And we need to support people to be able to understand what evidence exists about a tool's effectiveness, what doesn't exist, why it doesn't exist, and what to do about it. I think there's a really
12:34interesting role here for outcomes based contracting which I'm happy to geek out with anybody about after the session or during the session if anyone so chooses.
12:42Um and then the last is not unique to AI but I think is especially important and that's the stakeholder engagement. We have seen in the last several months a huge uptick in par parental concern about how AI is being used in schools but it is not monolithic. There's a wide variety of perspectives among parents and schools need to engage with those viewpoints and help bring parents along and stakeholders along and how AI is being used with their students. So that's all like when I think about AI revolutionary within the school system,
13:12I see a lot of barriers to that continuity of improvement, the scale um and whatever the third one is which I can't remember my own third oh changing of the underlying rules. So now shifting outside of the sector looking at schools like alpha like others that are AI first or AI enabled and especially I think interestingly in the micro schooling space and I know there's at least someone one person here from Prenda um and I a primer comes to mind as well as
13:39using AI as part of the school model and I think there's certainly some challenging of the norms right some changing of the norms I'm not so sure but challenging of the norms absolutely compress ing school learning into a short amount of time. Um, doing personalized learning at scale. Um, and I think that's something that we need to wrestle with what that looks like within the system, what is possible outside of
14:06the system, and what is it that makes it possible in one but not the other.
14:11Um, and then I think the um the things that are most in mind for me, and I'll I'll start being a little bit faster here. I feel like I'm taking up a lot of his time, but one is like we see personalized learning hap like trying trying to be implemented in the public school system and we're struggling we've been struggling with that with for a long time. Competency based learning, database instruction, differentiated instruction within a single classroom where students have wide range of abilities. Um data integration has been
14:39a long-standing challenge. um just because whether or not a student can really learn 2x in two hours is is a question for the researchers and that doesn't mean it should take till 8th grade for someone to reach third grade proficiency. So how do we find the space in between that and understand what can we translate from one from outside the system to inside of it. Um and then the last thing I would love to um have a conversation with the panelists about invite you all into is this question of
15:08student motivation. In an age of AI learning where cheating and shortcutting the learning process is going to become so easy for students to do. How do we actually ignite their desire to learn?
15:20So they don't want to cheat. They don't want to offload the learning. They want to engage. And I feel like that if any if nothing else is like the kernel of what we need to solve for as we move into this next phase of the AI the world of AI. [applause] Two both very tough acts to follow here.
15:44But what I will say is that my work takes me into communities across the country. So, I get to hear from school leaders, get to hear from um the civic leaders that are making some of these policy changes um and from a lot of parents. And I also, which maybe shows our age, took the prompt quite literally. And so I was like, "Oh, revolutionary rhetoric." Well, depends on where you're standing as far as I'm concerned. Like, we're not challenging the rules.
16:11[laughter] I went through traditional public school and so I'm going to follow the prompt.
16:16Um [laughter] but but here's the thing like we we know shamefully that state to state, city to city, district to district and classroom to classroom, the outcomes and the opportunities and the the challenges um are very different. And if if I can take a step back, I I want to bring us back to the early days of the pandemic when the digital divide, which we all kind of understood existed and was like a thing,
16:46became like a real exacerbator problem. Like we were able to see the kids who had the means to continue to do schooling and continue to excel. And we also got to see the kids who were getting lost in the cracks. And actually now that the social safety net that schools provided wasn't there to even try to catch them um what their what their outcomes were. And we're still seeing some of that um today.
17:14The printing press, tractors, the internet, like all of these have been real revolutionary disruptive pieces of technology. Um and yet they've all hinged on this question of privilege.
17:26And so like Rand did a study like not too long ago um where the wealthiest school districts have trained 70% of their teachers on AI and the poorest school districts have trained 27 28% of their teachers on AI. Um so there's there's this big question around the inequity that this creates that is unfortunately neither rhetorical nor revolutionary. It's just part for the course of what has always been like our our system. Um, so that's a thing that
17:55we need to contend with. Another thing that we need to I think contend with is the fact that like school administrators, policy makers, the opinion class, like we don't we don't see things necessarily beyond the headlines in the way that parents do.
18:10Um, we just participated in a a research study where we focus grouped a bunch of parents and I had my own thoughts about like how they're going to like look at this. Um, but what to me was I had all these pretty words as a comms person. I was ready to hit the like cable stations and just let you all have it. But like parents don't care about the R&D engine of public education. Parents don't care about all these like uh words. What they what they cared about was the
18:39flexibility um of their schools to adapt and help help their children meet the moment that they find themselves in. And I think you both mentioned Julian Kate judgment. That was the other big thing um that they talked about. Are these schools instilling in our kids the judgment to know when and how to use AI in a way that is conducive to their learning, not outsourcing it, that is safe, um and that is going to help them
19:07have the best time as 5-year-olds. I have two, so I really do hope that this is the best time to be a 5-year-old for my my kids' sake. Um but I also have to say that if parents are asking us for judgment on one side and I think that that is great. Um then you have school districts like New York or Los Angeles who are putting a moratorum on their AI development for a year for them to be able to exercise their own judgment on how this is going
19:37to be rolled out and applied in a way that is great and equitable and like it's going to lead to outcomes.
19:45um it feels to me like this is the the central question like judgment like how do we generate that and I don't believe that we create that or or engender that in kids by banning AI so that a committee of bureaucrats can decide how long it's going to take to do this and you know no offense to edtech but maybe take the lowest bid on a program or project that we don't know has any real outcome um because that's not how you learn to
20:14drive And I use that example because um there's a great uh charter school in in Newark called Northstart Academy, part of the uncommon um network that created this AI drivers license course analog cuz their whole thing is like you either learn to use it here in these walls and with this environment or you're going to go home and you're going to get on Snapchat and there's going to be an AI assistant or you're going to get on Meta and there's going to be all sorts of
20:43conflict. icting AI things or you're going to go order Domino's pizza and there's going to be an AI assistant there and like when people run out of claw credits they're now typing it into like the McDonald's AI assistant like hey can you like do this and it will give you answers just in case any of you are running out of credits by the way um so the the question to me becomes like are we going to like outsource our responsibility to help children learn
21:12and develop the judgment to engage with these tools that are going to be a part of their future um to the Domino's chatbot or are we going to do that in schools? I have a real clear um answer for myself um of of what I want to happen there. Um, and I think the other, so point one, the other point I made was around the chronically underserved and like how the disparity is going to
21:39continue to um, be perpetuated. And and we have to be real. There are people who will be able to exercise the choice to put their kids in a great school like Alpha or one of the others where they're going to become AI natives. There are people like myself who are now considering a forest school, would have never done that before. Um, and might exercise the choice of get my kids as far away from technology as possible for
22:06as long as possible in order to um provide them a different kind of um of approach to to bridging that gap. Um, we I believe cannot outsource either the responsibility to instill judgment or like the calling to meet the kids where they are and bring them up to the um the standard that the rest of their peers are meeting and that we um hold our you
22:35know to the promise of public education.
22:37Like that is that is our job and our responsibility. And there's a ton of examples of great places that are doing that um inside traditional districts or and from from my perspective like doing it really well through the charter model which allows for the flexibility while still having the calling to serve all kids because it is a public school. It is free. It is public. It is open to all and it has that mandate to serve all the
23:05kids. So, you're seeing it in Newark with Northstar, but you're also seeing it um in places like New Orleans, a kid school called Frederick Douglas um high school. It's a teacher called Emily Bissell who trained her model on her specific curriculum. Um mastery rates of the curriculum nearly like tripled um the highest AP history passing rates in the state of Louisiana. Um, and now
23:34they've been able to roll that to support 200 something teachers. Um, freeing up, let's take the the short end of the estimate, 4,000 hours of instruction that they're dedicating to teachers. Some other people would be um even more um optimistic and and call it 8,000 hours. Like we're talking about highdosese tutoring and how those things work, but like we wouldn't need any of that and any additional funding if we were getting kids the facetime with
24:03their teachers that they needed because they don't need to plan bus routes and they don't need to grade open-ended questions and they don't need to do all these um all these things. So to bring it back to the fine point, revolution or rhetoric, um I think that depends on where you're standing and I also believe that if we don't start to move then then it'll remain neither revolution or rhetoric but actually just another
24:31iteration of the status quo.
24:34Thank you all so much. [applause] I HAVE a few questions from all of that and then I'll open to you folks for some questions. Uh my first question is for Kate. Um you know AI has been around since the late '7s. The generative AI era was um roughly you know in 2022 was inaugurated. And so when we say AI we mean a lot of different things that are worth probably unpacking since you have direct student experience with your
25:02school of AI. AI has um in your words enabled um learning at 2x the rate of traditional students. I'm really curious what what AI AI are we talking about because if you were there eight years ago that that predates the generative AI era. I just don't want us to have kind of this fuzzy sense of AI magic tech dust that we're thinking about sprinkling over new school models. So could you be specific with for us here?
25:25For sure. So um you know uh 12 years ago um when um Alpha was uh founded the AI tools that the AI tools that the students were using in the morning were adaptive learning platforms. And so we were using tools like Alex for math. I don't know if you um have heard about that. Um uh we were using Khan Academy all of these pretty wide widespread
25:53widely known tools. Um and as AI has continued to develop and improve, we are now able to improve these tools. And so um we are now having students um uh we we are now creating tools that have um uh personalized uh reading comprehension passages for students. Um we are um uh students now use math academy for um our
26:23math programs in um uh middle and high school. And these the students are able to you know take an assessment and uh math academy will pinpoint exactly what it is that each student knows and doesn't know and provide them with the lessons that they need to do to um improve upon that. Um, and so, you know, AI is not at alpha a any sort of chatbot. And so, um, I think there
26:50that's a common misconception. Um, and the AI is actually just running in the back end. There's no sort of entering questions and receiving a response. It is fully um, ju used to create the programs and lesson plans and everything like that. of course audited by um uh our whole team of learning scientists and academic experts um but the AI the students are not engaging directly with the chatbot and so I think that's also
27:18an important distinction for um what AI is like in uh different types of academic contexts. So it has I guess short answer it started you know adaptive learning programs and now has transformed into AI powered uh te tools and um technologies for our students.
27:38Really helpful. Thank you Kate. Um so Julie I'm hoping that we can help one another you and me identify where personalized learning is on the the spectrum from rhetoric to revolution.
27:49I'll just kind of disclose my my read of the landscape. I I definitely get the appeal of personalized learning, a a promise that has existed for decades since I was a teacher myself 20 years ago. Um, one that promises to amplify human capacity to to spread teachers where they where they are most needed and enable kids to have exactly the help that they need when they need it. The promise is, believe me, fell at bone depth by me. Um, what I'm what I'm thinking about is the the research coming out right now, not enough for a meta analysis. It's still pretty
28:18threadbear, but the research um so far has not been encouraging to me and I'm curious how how you read it. Um one is the research from uh you know from the from the pregenai era around and zern and none of this is to um discredit like the work that your classmates a smaller group of students were experiencing.
28:36Lawrence Holt has uh his essay about the 5% problem which indicated that the the great results that are promoted by these companies on their their platforms come from the 5% of kids who engage in them at a certain level of dosage a 5% of kids which we might hypothesize um could be supported by a ham sandwich. I don't know like they're a certain kind of kid.
28:58And then the Gen AI era now we see uh research from Karly Robinson at Stanford showing that kids just like aren't feeling the chatbot programs like they they Carly Robinson put human tutors like sat them on these kids on this platform and saw you know usage go up by like a bare percent or two or in Tennessee most recently kids were just like it's a lot of IDK it's a lot of like you know like 67 into the chatbot window like it's it's not they did a real qualitative research indicating
29:27there wasn't a whole lot going on there. So, I'm hope you have a survey of the landscape at Bellweather. Um, you've got a lot about like where are we between rhetoric and revolution. I'm hoping you can help me and maybe others uh with their unbelief at the moment.
29:40Um, we'll see. I think um I think the um the I think just like we have to define AI, we have to define what we mean by personalized learning because I think there's a version of personalized learning that people think of as being um more of the the adaptive technology that makes sure that a student is able to move through the content and the competencies um at their own pace. And that's part of
30:08it. But I do think that we sometimes underestimate the importance of the other aspects of an educational environment that can be personalized to individual students. Um that might entail um a um the relationship with the teacher.
30:26It might entail projects out in the community that resonate with a student and help them connect what they're learning to the community that they live in and the problems they feel passionate about solving. Um it might include um you know themes around teaching kids about this was this is something I read about alpha school online. You can tell me if it's accurate or not of like you know adapting the content to a student's particular interest. So learning about the history of Europe by tracing the Taylor Swift era tour like to different
30:54cities right like whatever it is like whatever the lens is that brings a student in to be able to engage with the content. So there's so many different elements of personalized learning that I think um often it focuses on that academic sequence of competencies but leaves out those like those softer pieces. And I think that's I think that's one of the things that is so interesting to me about um the
31:21sort of the opportunity sort of unbundle learning is to think about what do students where do students get what they need and and who do they get it from? And do they need to get everything they need from one provider or do they have the opportunity to think about what kinds of what kinds of um experiences within or beyond the school day that can support them to achieving to achieving their goals. And so I think about the
31:49micro school movement and I think about the the very small learning communities that they foster. Um I think about um um the idea like schools over the you know have acquired so many responsibilities over the last decade.
32:04What would it mean for schools to be really academic specialists and to sort of engage with other providers in the community sort of in a community schools type arrangement to meet a wide variety of other student needs. And then as this like optionality or unbundling in education becomes so much more expansive um how do we help families navigate that? How do we help a family say this is the profile of my learner of my child
32:33what I know they care about? How do I pair the academic personalization with the afterchool program, the music lesson, the art lesson, the sports team, whatever else that child needs to give them not just the academic learning that is suited to them, but the learning experience that supports them into growing into a full and and functional and happy human being. So, I'm not really answering the question about the what the research says in part because I think the research um
33:03can limit how we think about what it means to personalize and I think we need to consider that full gamut. Um, and there is a role I think for AI in in some of this because this this ecosystem of learning options that's emerging um is complex and it's dynamic and it's a lot for any family to try to make sense of and it's a lot for any leader to try to help guide a family um on how to make sense of it. And whenever you have a
33:30super messy data set that's super hard to stitch together, um, you know, I think there's a potential for AI tools that sort of can stitch all of that and u and help families make sense of it. So I think, you know, in terms of revolution or rhetoric, I think it's I would say it's still in the arisatilian sense of rhetoric and like about persuasion, not about like the negative connotation of of rhetoric. Um, and
33:59we're not at revolution, but I think to get to revolution, we have to expand the aperture for how we define it.
34:06Would it be all right if I just spoke about my personal experience really quickly? Okay. So, um, one of the things that I have learned both at through my eight years as an alpha student and I also just spent the last six months uh, working at Alpha um, as a guide. So um you know at Alpha we believe that uh the edtech and the AI piece of it all is only 10% of the solution to um you know these uh this ed whole thing of improving education. the rest of the
34:35other 90% is motivation. And so that is actually why I believe the relationship um at Alpha we call the the um teachers guides because their role is um a lot more focused on guiding the student to um you know their uh success in life skills and providing motivational and emotional support rather than directly teaching the material which is what the AI does. But that relationship with the guide is so important because it allows
35:03the students to um have that direct one-on-one motivation. So, you know, that 5% rule, of course, you know, if a student is just given a um given a lesson, no one is going to want to engage in that. And I can speak from experience. That is a absolute boring no. However, when you pair that lesson, say like a math academy lesson with say um a individualized motivational model
35:31or you know uh something that uh the AI tools allow the students to do is have afternoons to kind of pick what they are doing in the afternoon. So for myself it was life skills workshops and then in high school I started working on um an Alpha X project which was my newsletter amongst other other um other things. But also you know we have uh Texas sports academy. So all the kids who want to uh practice and train and um become excellent in sport they get their they can do their academics in the morning
36:00and spend their afternoons uh uh doing what they love which is you know that sport and that training. Um and it's the uh technology that frees up that time for them to do that. And so I do I completely I completely agree with everything that has been said thus far that you know it really is about the entire experience a personalized educational experience that um uh really actually uh allows for the outcomes to happen.
36:28I'd love to follow up Kate um on Alpha School and help us all get a bigger picture of what's going on and what what kinds of the what parts of the magic can we adopt for our own models here and certainly motivation is something that you folks have thought a lot about. Um, I think we we should raise to the group's attention that Alpha is a private school. It's selective admission and the tuition is top of market in every market it's in to my knowledge.
36:48Um, the school up the road for me in Oakland, California, I think is uh 65,000. I do I do believe that has some effect on a kid's motivation. If my parents were paying that for me to go to school, they might be invested in my outcomes. Um, but I want to ask instead, I think it's really exhilarating to see you folks um partner with researchers to explore different models. um the sports academy. You folks have also ventured into the public school system which is uh just a very it's it's you know it's uh not selective admission. It's not feebased. Um so Texas preparatory uh
37:18schools and unbound academy um a virtual charter in Arizona. Can you help us understand how how how is the model adapted? Are we seeing 2x learning in two hours there? How's it been going?
37:29Of course. So um first for the majority of the um uh the the best part about the alpha model is that you know as I was mentioning you can choose your afternoon. So, Texas Sports Academy um in Texas, as um was mentioned earlier today, uh we have um a lot of the uh the voucher system and so our students are able to use um uh the vouchers for Texas Sports Academy and it is priced at a much um and so they are pretty much um
37:58uh almost all of the tuition is covered by the the vouchers. And so um uh and similarly we have other uh programs like Montasurium which is a Montasauri focused school along with um uh uh NextGen which is GT uh GT and um a multitude of other um uh a multitude of other programs that are able to be priced lower because of um just different um afternoons essentially. Um
38:27and so for um any the um alpha model implemented in um other schools I um have been working at alpha so personally that's not u a huge area of expertise but I do um go back to what I was saying pre earlier that motivation is 90% of the of the solution and so um you know we can put in whatever AI tools that we want but if we have not figured out the right motivation if someone doesn't care about sports if they don't want to um
38:56you spend their afternoons building this great project. If they don't um have that um additional motivation then um you know it is much more challenging for um students to um actually uh engage with the program. Um and you know uh whether you know we have parents uh answering all the questions for the kids um chat uh chat bots like all of that of is of course hap will of course happen
39:24if the student is not motivated and engaged. And so I think that solving that problem as um you know Julie was talking about earlier um is really the key to um figuring out um you know how to navigate AI in the um you know uh in the landscape of education.
39:42Thanks Kate. Can I ask you please go ahead.
39:45I think you you both said something super important. So this uh set of focus groups that I referenced earlier like one of the things that was interesting because it is not new is personalized learning. And so we were asking parents about that and the level of skepticism that people had towards personalized learning was staggering. Um and I I think that that is because none of these ideas are news. Like Waldorf has been calling teachers guides for god knows
40:14how long. like Montasauri's been around for a hundred years. Let the kid do whatever like they want and like be motivated and what have you. Like none of this is revolutionary. Like the technology is great. It's super helpful.
40:28I love the way that Alpha is um infusing it as opposed to just having a a chatbot. Um but at the end of the day, you talked about a little bit about the gap in the research and how it's thin and how we need to figure that out yet.
40:42the chasm that exists between parents expectations and the the things that schools are delivering or the ways in which they're failing to deliver what they promised to kids is huge and if we don't spend time there then I think AI schools are going to go the same way of like Montasauri and Waldorf and insert whatever program great niche niche thing that some people are going to pay top of market for um and other people are going
41:10to get some watered down version Thank you all for painting a really clear uh it's a complex picture. It's murky and uh but you're all exploring different facets of it. I really appreciate that. I'm going to pass it to you folks for questions. I'm sure you have many. I only beg of you in advance to please state your name and ask a question.
41:31Um do we have a microphone for this one?
41:33Great. Behind you, nearest to you, please. Uh uh us in front of you. That's fine. We'll get to you next. Go.
41:37Thank you very much. I'm Jack Panel, the founder of a school in Arizona that uses AI every single day. I um I transcribed part of this conversation with Whisper Flow and then I put into if you know what Whisper Flow is, I put the transcript into AI to ask what do you think is the most important question I can ask? [laughter] And I think that's an example I think of for me AI is a thinking partner with me. Uh, and it knows my brain kind of because I've
42:07inputed my whole brain into it. My question is, um, or I'll give you the question the AI wants to ask you. My my AI is called Icorus.
42:18And Icarus says, by the way, what is the evidence that AI itself is producing the learning gains?
42:29And, okay, that's not bad. That's not bad.
42:30That's [laughter] a good question. Well, I I gota I got to hand it to AI now and then. All right. So uh did AI say uh who should have that question? [laughter] I think anyone on the table and it specifically it it emphasized if 10% of the day of of the day is spent in AI then why is it necessary to have AI? I mean you specifically say AI is used in Beast Academy I think you mentioned or you mentioned math you mentioned you mentioned it it enables learning but
43:00it's not necessarily teaching learning.
43:03And there's a difference there. And I think that's where the claim needs to be made. I I I want to see what is the evidence that the AI is producing such this great outcome for kids. And why why isn't that part of the 90% of the day or why is only part of 10% of the day and maybe you don't need it?
43:20Um yeah. So uh to start uh I should have clarified a little bit better. Um the it's 10% of the solution um uh not just 10% of the day. And so um students are using um the uh time back timeback is what we call the software are you is use students are using the timeback software for 2 hours every single day in high school 3 hours for the AP courses. And so, um, I guess the the thing that the
43:49way that I would respond to that is that I the AI is not AI itself is not um, you know, uncovering these new methods of instruction. um what the AI is doing is allowing each student to learn at exactly, you know, the right place, exactly the right um uh at at their own pace. And so it's not that the AI is
44:16teaching better than a teacher. Um, the AI is allowing the student to uh learn the content at their own pace and thus ha freeing up time um to then go about um the rest of their the rest of their day and learning all of these different life skills and a multitude of other things. Um but I I guess it's not it's not that the I don't know if I I'm explaining this right, but it's not that
44:45the AI itself is responsible for the learning outcomes. the AI is just enabling the learning outcomes to take place um in two hours in the morning. If if that it totally makes sense and I think if you go back to the example that I shared and I have another one in San Antonio there's a school that's focused on kids with dyslexia that's also the kids are now um reading at grade level which kids with dyslexia that is tough um to do.
45:12It's not because the AI chatbot whatever taught them how to do it. It's because it allowed the teachers to create like real individualized learning plans for each of their students based on their specific shade of dyslexia which has allowed that teacher to then support that student appropriately. Same with the example in New Orleans and their astonishing passing rates in the AP test. It's not because they were learning US history, although we could
45:41all use some of that. Um, but it's because they were learning in a way that was meant for them, right? Like the it was looking at their open-ended questions and figuring out, oh, you need more information to understand like the socioeconomic components of the Civil War.
45:57So, just one follow.
45:58Well, one one second and we'll leave that to the moderator to decide the follow-ups. But I think that the other part of your question about the research piece is like thank god we're here at Harvard like premier research institution and that you know Paul is here like maybe they can do some um some studying and some researching about whether AI is responsible for the academic gains or not. But what we know so far today is that AI infrastructure plugged into schools gives teachers more of what parents want which is the
46:28flexibility to respond to their kids needs.
46:31Let's go in the very back please. very back behind you behind. There we go.
46:35Thank you. Uh Steven Wilson from Pioneer Institute. Kate, thank you for being here. It's very unusual to have a student from one of the models actually describing your experience. And look, I don't doubt that for a motiv motivated autodidact like yourself that the school would be tremendously successful, right?
46:56and for others who can afford the $40 to $70,000 tuition and the selection selection effects of admission. But what I wanted to ask you about is what you know about when the model is applied to ordinary children as at uplift education that was mentioned because the promise of your founder is that it will work for all public school students and they will achieve 2.6 times the rate of learning.
47:23Right? But in this particular school, which is a charter school that you mentioned in Arizona, Unbound Academy, uh the organization projected that 60% of students would be proficient at the end of the year in math, 10% were. They projected that 65% of students would proficient would be proficient in ELA, 28% were. So I guess my question to you is are your leaders at alpha
47:52overpromising and is there a risk that alpha will become the therronos of education?
48:01So um obviously um [laughter] thank you. Thank you. Um, so, uh, to start with the, um, Unbound, uh, stuff specifically, I, um, you know, I was the guide in New York and so, um, that's I'm not super familiar with the specific details. However, from what I do know, again, it goes back to the motivation piece. um it was an online school and so I think that it what what Alpha and you
48:30know the um uh administration is working with on right now is how to solve the motivation problem for all students um even students who aren't able to go to a physical school and so I think that is um you know that's uh was after I I one year I believe and um you know uh they're still workshopping and trying to figure out exactly how to solve um the motivation problem in that um specific instance. And so um there's that piece.
48:58The second um of course uh you know I I'm not sure exactly what um McKenzie has said um but uh I believe that there are the the best part about having all of these different models is that students can choose um you know and families can choose what works best for them. And so, you know, I if uh if for your family it would work better um uh uh the the more traditional model or um you know, any of the other models that
49:26have been mentioned recently uh work best then, you know, I um I I'm very happy for that. I encourage that.
49:34However, if um for students who um you know are um you know, Alpha isn't the the school for every single student, but it is the school for a lot of students. Um and I believe that um you know uh again uh Texas Sports Academy and um a bunch of the other new newer alpha models are not at that price point. And um you know students come into alpha starting at you
50:02know the fifth percentile uh achievement wise and uh they come in starting at the fifth percentile and um achieve that and are uh majority able to achieve that 2.67 six, seven, whatever the um number was uh uh that learning growth. And so um I think there is a um there's a um difference between um a school a student uh you know coming in with some
50:31um you know if a high schooler comes in and they don't want to be there, it's not going to work because you need that motivated student. Um, but that's the guide's job to create and foster an environment to help motivate that student. Um, and so kind of to wrap everything up, yes, um, uh, of course, alpha will not be 100% effective for 100% of students. However, um, it is to make it effective, we just have to solve the motivation problem for every group
50:59of students. And so that's Yeah.
51:01Could I jump in for just a sec?
51:03Yeah, I'd love to, um, direct attention away from Kate for a second. Give Kate a break. Um, so Julie can hop in and other questions for Julie or Steven would be appreciated here. Yeah, I mean I think that I think the question is the right question given some of the marketing and the messaging around Alpha School, right? There are a lot of claims about what it can do and for whom. And when I was think when I saw that you were on this panel and I've been following alpha school and I was
51:30thinking about this these criteria for revolution it struck me that alpha school might get to that discontinuous improvement that real change in learning but it is not changing the underlying rules of the system and it is not reaching scale. So maybe it's not revolutionary.
51:52Is it okay that it's not revolutionary?
51:54Absolutely. It's an experiment.
51:57We as a sector need to be learning from people who are thinking outside the box and trying new things in tiny little ways and we need to learn from those and see what does and does not apply. So I I I think like I feel quite confident in saying like revolutionary like not no like not yet.
52:16We'll see. But I also don't want to like throw the baby out with the bathwater and say therefore no good. Like I I think that there's I think we need to experiment in this space and we have to be comfortable with that and um the lessons we might take might be of a smaller grain size as a result those little tiny learnings but I think we need to appreciate those for what they are and we shouldn't take our our eye off the ball like it's not absolute proficiency gains that are the only
52:46things that matter. Like to Kate's point, like if a kid comes in seventh grade at third grade reading level and that teacher is able to move them up one, two, three grade levels over the course of that year, forget good. That's amazing. So like we also have to ground ourselves in reality because I think AI and all of its promises um have led us to believe that it's like the silver bullet that's going to like fix everything in one go from one day to
53:14the next. And like I've got a beachfront property to sell y'all like somewhere in the Midwest.
53:21It's right here, please. Yeah. You have a question for uh for Julie or Stephen?
53:27My name is Katie. I'm head of learning from Prenda. Um at Prenda, 40% of our kids have learning disabilities and our tuition is about $7,000. We operate mostly in state school choice funded operations where parents pay nothing out of pocket. Um so my question is one of the things that I mean we just started a hundred schools this year and one of the reasons that people say no to prena is that they they learn that we use screens
53:55and we have this huge backlash who it largely comes from the anxious generation and Jonathan height who is one of my favorite people and thinkers honestly but I'm like I usually am kind of on the same team as this guy [laughter] um and I am when it comes to social media and phones but I'm like Oh, we actually disagree on edtech, which is great. Um, so how do we guarantee that we don't throw the baby out with the bathwater and allow the power of edtech?
54:22It's just so such a so ironic that when this backlash is coming at the exact moment that AI is hitting schools, right? Um, so how can we create school environments, not outside the system, but inside the system, um, that support responsible AI use and to get some of these outsized results for kids that are just going to their neighborhood schools.
54:50[laughter] I'll do my best. Um I mean I think so one is um you know Dan's question from earlier is what do we mean by AI applies to this broader conversation about the techlash? Screens are not a tech. A tech is not AI. Like these are very different things but they're all getting pulled into this general very like um passionate um push back against how students are engaging with any kind of
55:18technology in schools. Um, and so I think we're we're we're in a moment where the the pedagogical is being trumped by the political. Um, we are in a political moment now and we are going to have to fight this out to some degree in little microcosms of democracy all ac across the country of what is or is not an acceptable use and how we bring different stakeholders in on that conversation. I think I you know I would say I really see like four pillars of
55:48this. One is we have to support AI literacy for students, educators and system leaders. Um so that they know they can describe what AI is and is not. They can describe the difference in edtech tools and are can be informed interlocutors with the people who are coming at them with questions.
56:06Um and they know how to use the tools responsibly as well. Um I think we need um states down to districts down to schools down to classrooms that chain of implementation support with policies that are actionable and specific about the ways in which AI can be used safely.
56:24what is like an Aussie uh down in in Washington DC um just released guidance a couple of weeks ago on their AI policy which is very hopeful red yellow green like don't do this you can totally do this before you do this think twice we need those kinds of actionable guidance for people at all levels of the system um I think we need some I again like to beat the procurement drum I think we really need to see procurement as a major place of intervention within the
56:52system to to shape the tools that get in front of students and how they're used and the kinds of results we expect from them. Um, and then the last piece really does come down to this stakeholder engagement. And we've like we've been on this path before in the sector where we have we have tried to make a big change but we have not brought the stakeholders along and we've tried to go too far too fast and if we need to slow our pace to have durable change I think we should.
57:20Thank you. Our our last question in the back please.
57:25Hi everyone. Uh my name is Joyce Joyce, a former principal in uh the Dallas area and a current student in education policy here. Uh my question really for anyone on the panel is uh hearing all the conversation on AI. I'm hearing that there really are two factors. It's either the AI for teachers and the training who are doing the the work for students and then the AI that is placed in front of you know the students and the AI works directly with the students.
57:55What would be in your experience across or even to make case for evidence for districts on when is it best to use AI you know for the teachers so that the teachers gain the the skills and the knowledge to you know old old school way teach the students um versus AI you know being placed in front of the students um so that you know it is the AI interacting with the child. Stephen, can I pass that to you to close? Thank you.
58:23Um, I I think of a couple of things. I think of something that Durell said in the first session, which is like this has been relatively around for a minute, so we need to give it five to figure it out, right? But we can't be um research and developing with kids either because that would be completely irresponsible. Which as a caveat is why you haven't seen any of the major AI companies like have a studentf facing like program themselves, right? like their technology powers. A
58:53bunch of other companies to do that because those companies are taking into account safeguards and a whole bunch of like particularities related to children that are super important both from a regulatory perspective and also just from like don't be a terrible human being perspective. Um, so my vote would be to first roll it out to the teachers because we know that there AI as infrastructure can have an incredibly powerful impact in the way that kids learn and the way that they engage with
59:22things from administration to curriculum development to grading and instruction like design. There's a whole world that we can attack in these five minutes that we're giving the technology to kind of figure itself out before you wholesale roll it out to kids without forgetting the innovation and experimentation and flexibility that is required to meet the moment as well and supporting some of
59:52the outliers like Alpha, like Pendra and like many others in the room to also get in front of students and try out how things are doing. So, not sure that it answers your question, but much like everything else in our space is a both and um situation here and um I'm glad that all of you are stuck in trying to figure this out because we we can't do it without all the stakeholders.
1:00:16A fantastic closing remarks. Thank you to our panel for their insight [applause] and contribution. Thank you all. How you doing? Good.
