"We are teaching to an average that doesn’t exist" — Robin Lake on reinventing public education

We have structures that we need to demolish, and I don’t mean demolish public education, but I do mean that we need to demolish the boxed set of assumptions that we have about kids. Kids do not fit into boxes. We are creating boxes, and we are instructing for an average that does not exist.

Both of those things are crazy. All right, Robin Lake, welcome to AIE D Studios. Alex’s in my bedroom, actually.

Alex’s bedroom edition. It’s funny to see you sort of the virtual screen having been with you a couple weeks ago now in in New Mexico. So, for Robin, you’re the director of the Center on Reinventing Public Education.

Small mission that that your center is trying to tackle. Hopefully, we won’t be in business forever, but we’re set up to be, aren’t we? Yeah, I suppose I mean I mean I mean maybe that’s like a So, reinventing public education is a like a big aim.

Can you Can you just say more about like what does that mean for someone who’s maybe this is the first time they’re they’re even hearing about CRPE. Or is it is it Serpy, by the way? Do we Should we Should I be saying Serpy?

Sure. It’s come to be that. Personally, I prefer CRPE, but nobody refers to us.

We don’t even refer to us as that, so Serpy is fine. Yeah, so it is confusing for people because reinventing almost implies sort of classroom-based curricular kind of work, which we’re interested in, but actually the center was founded more than 30 years ago by Paul Hill, who had been working on big questions in public education, big evaluations, and trying to understand why great schools that work for all kids, but especially the most vulnerable populations, aren’t everywhere. Why they tend to be really isolated and not prevalent, not reaching the kids who need them most.

So, he wanted to work on the question of, you know, what does it take to take effective schools to scale, what stands in the way, what is it about the system that prevents schools from being coherent and innovative and using the most rigorous practices. And so, he set out with the set of systems questions and reinventing actually came out of the tradition of reinventing government. So, there were a lot of people writing in the ’80s about how government should function differently to promote better performance at the unit level.

And Paul wanted to take that idea and say, "Well, what if we did reinventing government in education and we kind of changed the role of states and school districts so they were promoting performance and and giving schools more autonomy to do what they needed to do on behalf of kids." So, that was that was the origin of reinventing and over the 30 years and and by the way, I’ve been at the center for almost all that time, so since I was 7 years old, it’s fantastic. But, we we set out to study all kinds of different approaches to reinventing. So, we studied a lot around public school choice and trying to understand what was working and wasn’t wasn’t working and say charter schools and that kind of thing.

But, we also studied accountability and finance and teacher quality questions. We always have loved to look at those system questions, but really ground them in the classroom. So, while we are really not curriculum people, we love to be in classrooms to understand what’s working what’s not.

Or nowadays, not even necessarily classrooms, but alternative learning environments. So, that’s us and so what, you know, why AI, you might ask. I I’ll just offer that the bridge to AI for us came actually before the pandemic when we started hearing a lot about the jobs of the future and the way that new emerging technologies might change the workforce, but also much broader civic education questions and a lot of things that we thought would profoundly affect the education system.

And we started worrying, as we always do, that good things that might arise from this would probably accrue to more advantaged kids unless we got ahead of the problem and started thinking about how we get the solutions to more historically marginalized populations and make sure that we approach the questions of scale and system questions really thoughtfully. And it’s with any question, whether it’s school choice or accountability, sort of like, well, these things can be what we make them. We need to design what we want and we just need to design toward the future.

Yeah, I think it was And so so so CRPE hosted the Think Forward Fellowship in New Mexico. And it was it was an interesting gathering because I’d never I really hadn’t attended anything quite like it where it was you had us working basically the entire time and like from 8:00 a.m. From dawn till dusk.

And there was actually very there were there maybe there was a panel. I can’t even remember if there was a single panel. It was almost entirely like working sessions, lots of sticky notes.

But you came armed having done a lot of research into how schools are responding to AI and and I’m curious if And if I recall correctly, you one of the things you called out was that you know, a lot of the schools are failing to prepare teachers. You know, like what’s like what’s the most alarming thing you discovered? Is there Is Is this sort of like a common Is there a common thread or was like every school different and yeah, maybe just like sum it sum it that that research into sort of like a a sound bite.

Yeah. Or a long It doesn’t have to be a bite, a sound meal maybe. You can You can talk as much as you want.

Sure. Yeah, we did pack a lot into 2 and 1/2 days, didn’t we? And we had such a great group.

We wanted to maximize the time. The The research that you’re talking about is coming from a bunch of different studies that we have underway. And And the way we’ve approached AI is we wanted to really as much as possible map the emerging picture and help people understand where we are at different points in time.

And so some of the things that we’ve been seeing through different different studies that we’re taking on is teachers are not getting much in the way of support from their districts. We also hear from districts by the way that they’re not getting much in the support from their states. So not getting much state guidance.

But teachers and principals say that they’re intrigued by AI often times but you know, have real fears about it but are just not getting a lot of support and we’ve seen that in lots of lots of studies. The thing that really caught our attention though was you know, back to my point about where the benefits will accrue if we let things just unfold naturally suburban districts are out ahead of rural and urban districts in terms of their support for teacher prep around AI. And we see in the data that students in suburban districts, same deal.

They’re you know, they’re getting more access, more opportunity to learn about AI, to learn AI literacy, etc. So we don’t like that trend. We think that something needs to be done about it. And you know, when we’re studying early adopter school districts, the ones that are way out in front, they’re saying they’re intrigued, but they have a lot of competing priorities, a lot of things that they’re that are, you know, vying for their attention.

And so, they’re not able to give it as much time as they’d like to. In general, early adopter districts are more at the tinkering stage of AI, I’d say, than transformation. And so, that’s worrisome to us, as well.

And I I’d say that the last bit that we’re starting to pick up in our research is we’re talking to people out in the field about what they’re observing, whether they’re support organizations or tech providers, is they’re saying that a lot of the AI attention in schools right now is going to what I would call point solutions. Maybe it’s saving teachers 15 minutes of time, or maybe it’s scaffolding some instruction for kids with special needs. There’s nothing wrong with any of that, but I think that there’s a feeling now in the field that we need to move from point solutions to transformational solutions and really think about things like whole school designs.

So, as you know, Alex, we spend a lot of time talking about all of those things, ways to bring more coherency to what we’re doing in AI, and a lot more widespread support. And, you know, in my view, this just gets back to the same issues we’ve always encountered in public education when we’ve tried to, you know, grab hold of a promising new initiative, approach it with caution, and, you know, and and proper evidence, but also scale the things that are working, or move toward more transformational tools for kids. Yeah, you I mean, you co-authored this report, and you talked about like the wicked opportunities of AI and it’s I feel like you can you can take that to mean a few different things.

I’m curious like what what do you see what are the wicked opportunities? Yeah, well that that term actually came out of our our first forum the forum that we did about a year ago where we were it was a lot you know, a year ago in AI is just an eternity isn’t it? And at that time the people that we brought together were really ringing their hands about all the things that they were worried about which are completely legitimate worries.

But somebody in the group said, you know, we’re going to make progress we need to focus on these wicked opportunities and I think there are so many and they’re under explored right now. One of them is around special education an area where we spend huge amount of money with kids who really really need support and have a lot of potential if we can deliver on that support, but we’re not doing it well. The outcomes in special education are really really atrocious.

And so we need to do better and I think that we’re just tapping the surface surface with AI about what we can do for kids with complex or unique learning needs. So I’m excited about what could happen there. Things like career pathways is another area where you know, a lot of people have interest right now and changing high school so it’s much more relevant.

It operates in terms of personal personalized pathways for kids and really helps them start orienting toward their futures whether you know, it’s college or some other next step as soon as they graduate. So there’s a lot that that I think AI could do there in terms of accelerating career pathway work. And gosh, I mean, you know, opportunities in terms of this changing way that schools and school districts operate in terms of their data infrastructures, the way that they pull data to be able to craft instructional strategies for kids.

I’ll throw out one more, which is one that’s on my mind a lot right now. We know a lot about how we can help struggling students in things like math and reading. And teachers are not getting enough evidence-based information about what to do kind of in the moment when they’re working with a child who’s presenting a really complicated behavior or academic need.

And I really think that AI could help a lot on that front, just delivering kind of closing the gap between scientifically based instructional strategies and practitioners in the moment. So, you know, many more things that we could get rolling on, but right now, very few of the tools that that we’re seeing in the field are getting to that level of shifts and possibility, wicked opportunities. Yeah, I mean, cuz when I first read the headline, I was expecting the wicked opportunities to be like wicked bad.

But you’ve actually identified so like there are those risks, but then there’s also these like wicked good opportunities. And yeah, I want to dig into special were trying to play on the idea of wicked problems. Yeah.

You know, that there are wicked problems in in education, could AI help solve them? Special education’s an interesting one to me because it’s I think the jury is legitimately still out in terms of, you know, should schools be like rushing to deploy an AI tool? I think it depends on like what AI tool and, you know, I my advice to administrators is often, you should be rushing.

You like you should be you should be anxious. Like, you’re you’re a to be concerned about being left behind. But, the answer to that isn’t necessarily just, you know, you know, scaling some widget that you know, and and and frankly, like, most of the companies that are selling tools right now, I don’t know if they’re going to be around in 5 years.

You know, I think I I think the AI bubble is who knows, but there you know, but there was there were lots of bubbles around whenever there’s a new transformational technology. Well, it’s also that when when we talk to people in schools, you know, principals and and district leaders, they’re really afraid that if they grab onto a tool right now that’s free, they’re going to be charged a lot for it later, and they’re not going to be able to get out of that commitment. So, yeah, I I mean, I believe me, I do not schools should be rushing into solutions right now.

My feeling is we need to get more urgent and more intentional about testing solutions for the problems that are most the biggest challenges in education right now. Rather than just letting the EdTech community kind of come up with ideas in isolation of what teachers and parents and students really need and want that could make a big difference. And I think the other the other issue is that you know, and and Alex, remember we talked about this a lot at the forum is you know, there’s a real question about whether we are asking the EdTech community to kind of make the most out of the schools and the school system that we have right now, or if we want to ask them to design for the future and say, "Hey, we want schools to look a lot different.

Right now, they’re not very joyful. They’re not very fun for kids. We’re having a hard time motivating kids even to come to school, but if they come to school, it’s hard to motivate them to get on to a a tool that’s going to maybe accelerate their learning, but that they just don’t, you know, they’re not engaged in school.

They don’t see the point. They don’t see the relevancy. So, the bigger problem is I think one that we all have to be grappling with.

And I thought it was really, really interesting at our gathering that people talked about that really forthrightly saying, you know, do we have do we need to make a choice between is, you know, kind of get juicing the current model as much as possible with AI tools or saying, "No, we’re not going to prop up a system that’s not working and we’re going to put everything in building for the future." I think where we landed as a group was actually we have to do both. We have to be ambidextrous here. We have to recognize that there are ways that our current system schools could work a lot better, but even if we maximized that possibility, that’s not going to be good enough.

So, we also have to work on the future. You’ve articulated this like very I think it’s almost fundamental. It helps me articulate our goal when we engage with schools.

Because on the one hand, we’re often trying to tell them like, "Don’t focus just on the tools." But we’re also not like against using AI. And I just want to come back to like I think special education’s a great example of this where should you be rushing to deploy AI tools broadly speaking? Maybe not, but and I want to talk about that, but there are some places where there maybe is such an obvious use case.

Special education strikes me as that where when you have students with like such a wide range of learning differences, many of which we’re really starting to understand more and more cogently than ever before. And and we and yet we’ve integrated special ed you know, special education. We integrated those students back into general classrooms without providing teachers with the requisite support.

And even though there is research, right, that shows, you know, the power of not segregating that group into sort of like a separate learning silo. But the research doesn’t support, you know, throwing them into a classroom and then not giving teachers that the the training and the support to actually, you know, provide differentiated learning and to those kids. And I don’t know that I don’t know if there’s a tool yet that exists.

It just feels like it could, right? Like it feels like you could build a teacher coach that is grounded. And maybe this is rag or maybe it’s like like retrieval augmented generation, which is like, you know, a database of, you know, vetted content that the model is referencing when it’s producing outputs.

So, you could imagine a tool that like has this sort of like repository you know, whatever it is that the that a specific district is sort of like using as its reference point whenever a teacher’s developing an IEP and coaching that teacher with, you know, ideas for how could you How could you modify a project to be more accessible to, you know, some of your students with those learning differences? I guess it’s not a very big market relative to just like {quote} an AI tutor, which might be why you’re not seeing as many, you know, venture backed AI tech companies. Maybe you’ve heard of some.

I don’t know if there’s like I mean, I should probably start with that. Like are are you aware of anything that maybe already exists that I’m just not, you know, clued in? But Yeah, I mean, I’ve talked to some companies that have been doing edtech tools for special education and are starting to integrate AI, but I don’t think that they’ve really figured out where the big shifts could come.

I think they’re making marginal use of AI right now. And most of them are still you know, as I say, kind of working within the special education box, right? You know, we’ve defined special education a particular way.

In my view, we really need to upend that model. It’s really not working. We’re throwing more and more kids into special education.

Many of whom probably don’t have a disability per se, but have fallen behind academically so far that we don’t have the tools in the general education classrooms to be able to support them. So, special ed becomes the solution. So, what I’d love to see is more companies thinking about how could we help the school really zero in on tiered interventions in really, really effective ways as a coherent model in schooling rather than saying, "Oh, there’s a child in special education and they require these services and so we’re going to do these things." How can we get more preventative in terms of special education supports and and really think about not just delivering the basic minimum services to kids, but really thinking about what is this child’s potential and how can we maximize it?

And Alex, I had a I had a son. But when he was in when he was in K-12 in the K-12 system, he was identified as special education. He was also what we call twice exceptional.

So, he had he had some giftedness as well. And nobody knew knew what to do with him and it was he just didn’t fit in the boxes and it was a disaster. It was really a disaster and we had to, you know, constantly advocate for him and we got him through.

He’s doing amazingly well. He’s thriving in college and grad school now. But, we need to do better in special education.

And so, that’s why I’m I’m urging folks to just, you know, kind of only think bigger and challenge all of our assumptions about what we’re doing with schooling right now and not to say that AI is the solution is going to be the silver bullet for solving all those things, but we have this new technology and we have other technologies and we have other potential solutions around rearranging staffing and things and so let’s get thinking along the the lines of the schools that we want to see, the solutions that we want to see and whether AI can or cannot play a role is is almost beside the point. Yeah, I I I’m generally skeptical of of think tanks. You know, I think there’s just there’s always so much like it’s so easy to get caught in the ivory tower and you know, as an organization that’s so focused on the grassroots, you know, we’re we try to avoid getting our head in the clouds and yet you are and and this is is also my takeaway from the the forum.

Actually zooming out and then and saying let’s talk about what what are we reinventing education towards? Like what are those goal posts? Cuz actually until you define where we’re going, it’s very like AI can get us to a lot of different like destinations.

And I think the state of play and even maybe in your report where you sort of identified right is this you know, rushing into AI without knowing where you’re going is actually not, you know, there’s there’s a lot of risk there. You need to figure this out and so this feels like the precisely the right moment for an organization like CRPE because I think the K-12 space sometimes suffers from and this may be not their fault that be because of the system they are not necessarily empowered to have imagination. Like Yeah.

I mean, is it possible for us to figure this out within the system we have or or do we really need to just sort of like completely change the system? I mean, how how how dramatic of a of a reinvention do you think is required? Well, Like right now, like in the next year?

I do think we have we have things that have to go. So, we have structures that we need to demolish. And I don’t mean demolish public education, but I do mean that we need to demolish the boxed set of assumptions that we have about kids.

Kids do not fit into boxes. We are creating boxes and we are instructing for an average that does not exist. Both of those things are crazy.

And so so there I’m pretty radical. There I say, we’ve got to change things up a lot. We have to think about really seeing every child as the X to solve for and making schools and learning systems flex around those needs.

That’s my fundamental belief. And that takes some creativity. It probably takes having educators work in teams.

People take on more specialized roles, but work in highly coordinated teams. We really leverage community assets and resources to be able to serve kids given their needs. We have to, you know, kind of act like this is an emergency because the results out there are not good.

The bottom’s falling out if you look at the data. And by that I mean the 10 and 25% lowest performing kids are continuing to decline in performance. And it’s really only the very, very top tier of students who are continuing to move along through the system at a stable rate.

We have big problems, and so we we’re going to have to figure out how to leverage evidence-based solutions at scale much more quickly and deliver solutions to kid and kids in much much more customized ways. Now, that said, I do not believe that schools have to look radically radically radically different in order to do that. I’ve seen schools that look fairly traditional when you walk in, but they’re doing all the things to meet kids’ needs in in really really customized and effective ways.

But, we we do have to get more nimble and adaptive, especially at the high school level, where kids have to be prepared for their futures much more quickly in a in a much more targeted way. We We just we can’t waste time. We shouldn’t have been wasting time previously, but we can’t waste time going forward, and I hope that none of that sounds like anybody is at fault here.

I really don’t believe so. I think we have a system and a set of structures that we set up you know, a century ago that haven’t been working for a long time, and with the jobs shifting as they are, with with kids’ needs shifting and becoming much more intense, we just have to we have to finally deal with it. We have to update our system.

I think something that that came up at the forum that you know, you you might be alluding to this like you know, teachers, administrators are are bound within the parameters that, you know, standardized testing establishes. And that seems to be coming up I think louder and louder, I would say, is this acknowledgement that, you know, until we really address, you know, what isn’t working with testing? It’s going to be really hard to think out of the box in the way that you’re describing.

And what would that look like for us to throw out standardized testing if that’s what actually needs to happen? I So, I believe in in testing. I think some form of standardized assessment in in terms of being able to you know, compare kids across the country on a common yardstick is really important because I mean, I’m a researcher, Alex.

I’m without some kind of common metrics, we can learn very little about what works and how kids are doing, who needs solutions most quickly, where our achievement gap is, whether it’s growing or not. Yes, it’s growing. And you know, doing something about it.

So, that said, I think assessments need to evolve a lot faster than they have been evolving. I need I think they need to really really important, they measure the right things. And you know, there’s growing evidence that what some refer to as durable skills or soft skills or whatever you want to call them, but the kinds of skills that employers say they are looking for.

The ability to communicate, the ability to adapt quickly and learn new things quickly, you know, all the ability to collaborate, have empathy for other humans. You know, there’s there’s pretty strong evidence that those things matter to employers, communication. And so, those are hard things to measure on assessments, but they are important.

I think we also need to get clear. Most of the things that I read say, yes, those durable soft skills are important. However, basic foundational knowledge also continues to be really important.

But maybe not all the things that we’re testing and all the things that our standards state standards say teachers need to be teaching are as relevant as they used to be. So, I really think it’s time for an audit of those things to say, you know, what are the things that we can agree with employers, with civic leaders, etc. Kids need to know and be able to do upon graduation and how will we know if they’re learning those things. And the last thing I say that a set where assessments need to evolve is assessments have been set up on grade level content, grade level standards, and we have a system where there’s every incentive to keep kids oriented around learning this year’s thing whether or not they learned last year’s thing or or 3 years ago’s thing.

And that’s a problem. That’s creating holes in knowledge that are causing kids to really struggle long-term in school. So, we need to find ways to have more mastery-based assessments so kids can demonstrate knowledge in this, demonstrate knowledge in that, and it’s stackable, and they can they can keep going, and if they have holes in their knowledge, we flag those quickly and they get addressed quickly.

So, we have a ways to go in assessments, but I do think that new technologies are going to allow us to they’re already starting to allow us to create new assessment items like on creativity and collaboration much more easily than we used to be able to do that. To make assessments less onerous for teachers and students so they can do things you can do the testing more quickly. You don’t have to 3 days on it or whatever.

So, we we’ve got to evolve them, but I don’t think that it’s time for them to go. But that nuance is helpful, right? It’s like cuz I think even for folks who are are are worried that standardized testing is you know, getting in the way of innovation in schools, they aren’t necessarily people who believe we shouldn’t be measuring.

Right. But what you’re doing is articulating like, okay, this is what it would look like for us to continue measuring, but maybe what we’re measuring shifts or how we’re measuring shifts. I I actually just before this I had an interview earlier this week with Jeff Riley.

He’s a former Commissioner of Education for the state of Massachusetts. He was he was describing when he was a teacher in Baltimore schools with the MSPAP. Are you familiar with MSPAP?

No. This is the the Maryland School Performance Assessment Program. And so it was for like I think 1993 until 2002. And it was like So, I’m just sort of paraphrasing from Jeff, but this is and I did a little bit of research after.

It’s like why they considered like one of the best standardized tests ever created. And what made it unique is that it was open-ended. Students worked on groups for a portion of the assessment.

And it was designed by teachers as opposed to commercial publishers. It was interdisciplinary and so students were demonstrating knowledge across subjects. And and you talk about like these like durable skills like communication, collaboration, critical thinking.

And it and it was actually really focused on those skills. And Jeff was like, this is the best standardized test in my view ever because you know, he’s like, I worked with students who, you know, were were falling behind and were often, you know, were were getting disciplinary, you know, remediation. And it was the one test they were excited about.

They actually wanted to go and take this test because it was just so engaging and interactive. It’s a casualty of No Child Left Behind. So, that was that you know, it doesn’t exist anymore.

But, you know, it struck me that while this feels like an intractable problem to some, like we actually do have I mean, you are you you shared a a vision for this. Like, it’s not impossible. Like, there is there is like there are folks who have an a sense of like what a path could look like.

So, like if an if a superintendent came to you and was like, "What do I what should I do in my district?" And I mean, paint the picture of a district cuz I’m sure they’re all different. So, let’s say this is a, you know, 20,000 students. This is like Lakota or that cuz it’s all two.

Well, 20,000 student district, it’s sort of like a blend of suburban, rural, you know, relatively diverse student population. You know, they don’t necessarily have, you know, computer like the like magnet they don’t have a magnet program or a bunch of computer science teachers. Maybe they have one who’s doing like a robotics after-school program.

And they’re still in a tinkering phase. You know, they’ve gotten their hand they’ve done some workshops. And he’s under pressure or she’s under pressure to come up with an AI strategy.

Yeah, what like what would your advice be to this person? I can even play the role if you want to role-play, but yeah, like what would you be what would be sort of your advice or what questions might you even ask to provide that advice? Gosh, I mean, I guess I pass on the advice I got from somebody else when I started thinking about how I could learn about AI and and whether what it might mean for my organization.

One of my friends who’s pretty tech-savvy said, "You need to become something of an AI expert. You need to read about this yourself. You need to play with the tools yourself.

And, you know, start getting pretty adept at understanding what’s real and what’s not. So, that’s where I started. And I think it’s it’s really important for any educational leader because you want to be able to sort of sort the the fluffy stuff from the real stuff, right?

Start building an understanding of that, but also there’s so much fear and sometimes resistance, from various parts of the educational community, from sometimes from teachers, sometimes from parents, rarely from students, I think, but, so but that change management process has to start with clarity about here’s the goal that we have for our system. This is where we’re trying to move in terms of our education system, what schools look like, how teaching and learning works. We’re looking at AI to see what it can do to help us with that.

And here’s what we’re seeing at this point. Don’t worry, we’re going to go into this carefully. We’re going to do some pilots.

I’d say the great thing to do is to start organizing students and educators to run some pilots around things. Say, we have some challenges in our districts the district that we’d love to solve. And, you know, hey, start start playing with this and thinking about what the existing tools might be able to do for us on this front, but also if you could invent a tool, what would it look like?

And, you know, there are organizations out there who can take, educators through that invention process if they want to do that. So, you know, I think playing right now is really important and and understanding what this thing is, what it isn’t at this point. And, but always to stay really, really grounded in what are our goals, what are we trying to do, and how could this help us.

And, of course, pay attention to the risks and the guardrails, and take them very seriously because, you know, the questions of privacy, the questions of whether kids are turning off their brains or turning on their brains when they’re using these kinds of tools and any others are really important. And so I think it’s it’s the right thing to do anyway, but it also helps calm people’s fears if they understand that their fears are are real and need to be taken taken seriously. Yeah, and that’s probably what they’re not getting from the tech companies, right?

Is is that there is not always going to be that because they just have different goals, right? Like an edtech company is if you’re venture-backed, you need to be posting year-over-year growth and 40% is not good enough. You need 100% 200% growth.

And you know, I think superintendents are like they’re whether they’re directly aware of that or they just can feel it you know, in the in the salesmanship. The suits that I talked to are inundated with people that want to talk to them about AI. But most of these folks are not necessarily interested in that superintendent getting on the learning journey themselves.

Which is something you said that’s like that I’ve been really pushing even my team on. I was like, you know, I think they’re just that some of these folks need to get hands-on. And we’ve gone through this question of like, well, what is the purpose?

Do we really think superintendents or even principals are they really going to be the ones building tools? I don’t know if that’s the case, but I think there’s I think to get to where you are which is sort of understanding the sort of the role of the tools, the role of you know, sort of the learning sciences and you know, education experts and administrators, you have to have a sense for what’s possible and for what the you know, what the capabilities are. Yeah, and to be able to sniff out the snake oil salesman.

I mean look, you know, edtech does not have the best track record in in providing kind of evidence-based solutions and being really transparent about what results they’re getting and whether and this particular tool is going to be worth the precious funds that the school district may allocate to them. You know, the evidence levels are just not super high right now, so it’s really worth taking a skeptical eye to the tools, but also holding out for optimism and insisting that EdTech developers are paying attention to the questions that districts are struggling with. So, if they’re not seeing the tool out there, they should be, you know, calling Google or calling somebody and saying, "Hey, we have this problem.

Would you help us solve it?" And what I see out there is many school districts who have done that have partnerships now. They’re allowing them to craft something that could be models for other places because they’re willing to just ask. Okay, so you have 12 minutes.

You have two options. This is a fork in the road. I actually have a lot more questions just sort of like nerding out a little bit more about AI adoption, even what could be coming next for like the Think Forward Fellowship and you know, coming out some stuff coming out of the forum.

We could also build a tool together. And I know I’m curious if you’ve actually done any vibe coding yourself. Most haven’t.

Even the Even the super Even the AI superusers, interestingly, haven’t necessarily vibe coded. Yeah. But if you’re up for it, we could actually build something.

I’d be curious what it would look like for us to build a tool. I mean, I’m very intrigued by that. I have no idea if we could do it, but I’m going to Okay, so I want you to play the role, you know, thinking back to your experience as a parent with a a student with learning differences, what What is something that you would like that that would have been amazing for you to have back then?

Like What are What are What are What are there any examples of I mean cuz I’m doing sort of the problem identification right now first. But like what are some of the things you struggled with that Ooh. I mean in terms of a tool, one thing that would have been amazing for me, it was a continual frustration was that the educators who were working with my son didn’t know as much as I did about how to meet his needs.

So, they didn’t know first of all what were the things that that were just good motivators for him. Even though they make you fill out all these charts and stuff, nobody has time to read them and, you know, and then and they’re not great charts. They didn’t know really like how to motivate him like how to get him to open up and engage.

And they didn’t understand that specific tools that they were using weren’t appropriate given his unique combination of things. So, for example, on that latter point, I was reading all of this stuff about educating kids who had, you know, a lot of intellectual ability but also had a lot of autism-like characteristics. And yeah, it’s just it’s a unique a set of circumstances and it doesn’t fit neatly into the instructional strategies a special ed teachers have.

So, like I would have loved to have a thing where some kind of tool that said, "You’ve got Benjamin. He has these characteristics. He’s really, really interested in how the world works and getting answers to questions.

He’s high anxiety and that makes him sometimes shut down if he can’t do something perfectly. You know, just these like here’s Ben and then here are the kinds of strategies given Ben that might be effective for him. Okay.

That I mean that theory basic tool just like here’s your unique Ben, the X to solve for, how can you adapt your plan for the day to be able to meet his needs. Okay, I’ve got some stuff cooking. For me?

Yeah, I can. The while while I got this thing cooking, are there any like maybe things that you that in terms of like materials, like research or frameworks that we could feed into the tool so that it’s referencing something like that’s like grounded in a source of truth? I mean, I I read a bunch of books on kids who are twice exceptional.

There’s a whole literature out there on them. I don’t know. Can you share some of your favorites?

Maybe we can just start with like one or two. Oh, I can’t remember the name. It’s been a while.

A search. I’ll do a search. He’s trying for now, so Okay, so let me just share my screen.

I’m going to show you what I’ve been doing. Google Gemini. Okay.

So I kind of just captured roughly what you said. I changed names. And I basically was asking like, okay, I need an idea for a tool.

And so here’s what the here’s the tool that the that that Gemini and this isn’t actually live coding yet. This is still sort of like designing. So like the specific combination of traits that requires a new one strategy.

Okay, so like the cheat sheet generator. So I I think that’s kind of what you’re describing, right? It’s like parent fills out some sort of structured form.

The tool generates a focus like visual PDF or mobile-friendly link. And it includes some if-then logic. How does this sound for you?

Sounds great. You know, Alex, it reminds me once upon a time I visited a school in New York City, a school for autism. And the teachers there wore iPads on their hip.

And every time they had a challenging situation with a kid, they could go to their iPad. And I think I don’t know all that was in there, but it was like information about each child in the school. And then there was a set of tools that they could tap into.

This is what I’ve been dreaming of. So something like a follow-up questions. Yeah.

Do you want the tool to be like, you know, a simple text formatter, like you know, should it just like to the parent be inputting like the literal like here’s the instructions for my student? Or should it sort of translate into sort of like teacher speak? Hm, I mean my guess is teacher speak would be ideal.

Okay. Yeah, a cheat sheet can be like everything or or nothing and you know, a teacher can only scan a couple things. So if they’re only looking at the document for just like, you know, like just like a quick glance.

What which category do you think is most important? So like the safety triggers or like the de-escalation or is it like actually like the like supporting learning and curiosity which is actually you led with, but Yeah. Gemini seems to want to start with the safety trigger.

Curiosity because I mean, I think most parents are like me. They want their child to be seen as an asset and they want teachers to be able to see their potential and and tap into ways to get to that potential. Okay.

I think we can skip that. Okay, I think this is enough. Let’s let it Let’s let it run there.

I’m going to do another I’m going to do a I’m going to do a Oh, Gemini’s running, so I guess I’ll use Claude for this. Let’s do a deep research query. Can you compile a all the Has the research changed that much in the last, let’s say, decade?

I mean, how far back would you want to look if you were going to compile, let’s say, like you know, sort of a definitive overview of like the available research that we’d want to reference? I’d say two decades is probably good. And how how how focused do you think this could work with students with learning differences?

Should we focus on like neurodiversity or even just like autism specifically? I would suggest starting with autism specifically and maybe building out from there. Autism has enough complexities and it when it it if you can put autism up against giftedness and you know, interact autism with with some of the you know, a few other characteristics, that could be interesting.

Eventually, I’d love it if you know, if a tool like this could have not just diagnoses, but you know, really zero in on characteristics across diagnoses. Cuz to be honest, my son had an autism diagnosis. I don’t think at this point I don’t really think he’s on the spectrum.

Maybe. But when we got that diagnosis, the the person who gave it said, "Like I’m giving him a label cuz that’s what you need to get services, but he’s not a garden variety anything." And that’s true for a lot of kids. Yeah, and I I So there’s actually it’s funny, there’s a company that actually is building something like this.

It’s it’s more for younger kids. And I’ll I’ll put the name in the description because I actually forgot. But it was it was founded by a parent of a of a child with autism.

And it literally was this exact problem. He was like, "I I encounter a lot of situations throughout my like my life outside of school where it would just be helpful for folks to like understand my child. And I find myself constantly kind of like explaining that.

I just wanted to be able to like generate like these cheat sheets. And I believe what it does is it’s it’ll generate the cheat sheet based on the persona of the person. So if it’s like, you know, a caregiver or like a nanny, it won’t be in teacher speak, it’ll be in you know, just sort of more colloquial.

If it’s like a therapist, it’ll be a little bit more like it’ll draw like more terminology from psychology. I can’t tell you how many letters I had to write to other parents at the school trying to explain, you know, "This is my son. These are his behaviors.

This is why he acts like he does." that would be a great that would would a great gift to have something just, you know, kind of make drafts of those kinds of things. So, this is This is vibe coding. This is a an app called AI studio.

If you heard of Replit or Lovable or Cursor or Claude code, sort of a variation of that. I think mostly works similarly, right? So, you have your your text chat here.

This is where your like sort of like, you know, semantic coding your vibes come in. And on the right here, you have your preview screen. So, in short order, we should have sort of the initial mock-up of an app.

And this is actually like a like truly like a web application. I can I can share the link with you and I will, so you can sort of play around with it after if you’d like. I can also just directly publish it so that it becomes a live website link that people could use.

But hopefully what you saw, you know, I think what’s And you know, you’ve you’ve you’ve been around the block a few times, but even for someone who has no background, right? Like I I started with you know, literally just listening to what you were saying. And I didn’t The only thing I provide The only thing I said was like, you know, "Can you I want an idea.

Like help me come up with the idea." So, I didn’t even have the idea. And I have a sense of how to structure a prompt for AI studio, but I didn’t assume, you know, I assume that this is someone that who has no idea how to how to get there ask some follow-up questions, and it actually gave me the literal prompt that I just copied and pasted, which is what you see here. Okay.

Let’s start with Jimmy. And it always does this with the black text. This is a great example.

It’s the begin profile. All right, I’ll let you I’ll let you I’ll let you drive. Which one which Where would Where Which one would you start with?

Cuz there’s three different options, I think, that we could start. Let’s do the interview. Okay.

Hey. Like what are some things that you mentioned like just curiosity about the world or is there maybe a more specific example of like something where they just like, you know, where they just like really just sort of nerd out and go deep and get get super engaged? Oh, yeah.

Anything about how the world works. What happens when you know, deep deep philosophical questions like what happens when people die. I got that one a lot.

But yeah, it’s something that’s really intellectually engaging. Is what made him lose track of time. Also music.

Like making music or just listening to music? Anything about music. All right, next question.

Oh yeah, that’s what when do they sort of shut down or get frustrated? Perfectionism gets is is a problem so would make him shut down. As in somebody with high standards or Yeah, making a making a mistake, losing a game, anything.

He could he could imagine something that was perfect and anything less than that would be extremely frustrating and cause a shut down. What’s an environment where they thrive? Like do they need a quiet corner with headphones or do they like working on the floor being outside?

Reading books, quiet corner, space, physical space, kindness and patience, flexibility. You mentioned music. Mhm.

And what’s their superpower? Music. Music and, thinking deeply about things.

As well as, deep empathy, actually. Oh, and so, do you know what’s here? I I’m going to make two edits, for starting screen.

Text box is black. Make it white, in interview questions, enable user to go back and change their answer instead of starting over. Is there anything else the teacher must know?

Nothing for now. So, what we don’t have here is, let’s see what let’s see the status of our deep research. Ask them follow-up questions, dang it.

Create, all of K-12. No. to support teachers. Yeah, sometimes I forget that it’s It’s good that it’s asking those follow-up questions, but Oh, wait, we have our I don’t love the formatting.

Mhm. This format is crazy. Yeah, but I mean, that’s that’s, it’s pretty close to what I would have recommended to a teacher.

But more. Yeah, this is And so this is where the this is where like, you know, it’s you’re never going to get the perfect This is truly like the the most rough cut of a prototype. Right.

I think what I would want to feed in is I would be I would want to be thoughtful about like, you know, like sort of like any research-backed approaches that we could like use as reference points so that it’s like even if it’s not directly citing them, it is grounded in some sorts of truth. Yeah. What else would I want to add?

And some teachers might appreciate some really specific tools, even worksheets, or lesson plans or something on a particular item. Yep. Can you include a section for specific tools, worksheets, or lesson plans on a specific item.

And please fix the formatting in the All right. Anyways, obviously this is like this is going to be a work in progress. I will Let me share with you so that you can if you if you feel like continuing to riff on it.

Thanks. Yeah, that’ll be fun. All you need is a is a Google account.

I I think you Everything you need to jump in. So just to And maybe just to close things out, you know, you the one thing you said that really struck out stuck out with me is at decision-makers in education need to do their own learning. Like they need like Cuz one of the things we talked about in the forum is you know, how do we provide benchmarks or like a framework for evaluating tools.

Which I think we need that for sure. But I I I’m with you that those they will never be sufficient without a savvy administrator who knows what questions to ask. Yeah.

And I don’t think the framework is the way to get them to be able to ask those like cutting questions. I think it’s just enough comfort to you know, here’s something and be like, "Ah, is that actually you know, Right. Yeah, couldn’t I do that using, you know, these two coding programs that I just saw?" Ooh, hold on.

I mean It fixed some stuff, too. Oh, let me reassure you. Let Yeah, I mean, let’s not blow this up opportunity, right?

Like we we can’t afford to it’s too important to get it right and potential’s too great to miss the opportunity. And so, yeah, we all need to get a savvy and help each other. The the thing that’s been comforting to me is there are very few true AI experts out there.

So, it’s not that hard to become, you know, I wouldn’t call myself an AI expert by any means, but means but it’s not that hard to become, you know, pretty comfortable with it and as up to speed as almost anybody. So, it’s worth taking the time, but you know, I think it’s also right not to get distracted by the bells and whistles and to stay really, really focused on what’s the educational goal, where are we trying to take teaching and learning, and what are the tools that exist right now that could help us? If they don’t exist right now, who do I call to start working on that tool?

Yeah. It’s kind of fun. I mean, what we just did was fun, right?

Like the opportunity is really great out there to say, "What if what if we could turn special education on its head and get radically better outcomes for kids with disabilities?" And I could imagine like that experience even ending with, "You know what? I think we actually have something like this already. I don’t know if we’re using it in the way that we could." Mhm.

Like I don’t know if the answer is always like build a tool. Right. But I think sometimes just that like it’s almost like the design thinking process, which is where I think AI you know, I don’t think AI is like going to replace teachers, but it it does have so much potential as a learning experience as a as a thing that generates learning experiences.

And so there is something meta about like getting administrators and the and teachers, right? Like to you know, like walk the walk, you know, before you it like the idea that teachers are going to teach kids about AI literacy sometimes is I roll my eyes because like honestly the kids in some cases are way more hands-on. They’re the ones that have who have been running with it.

And that’s why they’re getting away with cheating, by the way, cuz they’re just like much more savvy at chatting their assignments. Yep. And and there’s like this and you know, technologists have this habit of making tech feel like hard you know, like a like creating barriers to entry just because you know, using lingo and things like tokens and inference and and rag and things you know, the even words like language that I’ve used.

But the you know, the the entry the entry point is lower and there’s a specific expertise that you’re naming, right? Among educators and the education space that can’t be that the techies don’t have, right? Yeah.

Yep. Robin, anything else that like for just someone who’s listening and they just want to nerd out some more should they just go to Surpi’s website where where Yeah, we have so much stuff on our website and yeah, crpe.org. You can find a lot of the data that I referenced, but also track these reports that we’re putting out from the fellowship.

But, you know, the people like you, Alex, that we brought together are the real resource and we’re just, you know, bringing people together to share insights across them and to be able to synthesize and get that back out to the field. So, we’re trying to do that kind of sense-making for folks. It’s just such an overwhelming space.

I also run a Substack if anybody wants to follow that, they can find Robin Lake Think Forward Substack. And yeah, just stay in touch. Let us know what’s going on.

We’re always looking for bright spots, too. So, Alex, if you or you know, others are hearing about things that we should know about, please let us know. We’d love to just lift up great stuff that’s happening out there.

All right, Robin. Thank you so much. Thank you.

All right, talk soon.