Enhancing curriculum with strategic AI in education: Practical approaches with Adam Garry

Rich Henderson:

Welcome back to another episode of the Leading and Learning with AI podcast. I'm your host for today, Rich Henderson, Vice President, AI Solution Architect here at Panorama Education. I am joined today by Adam Gary, who has over 2 decades of experience consulting educational institutions on their journey toward digital and student-centered learning transformations. Over the last 20 years, Adam's contributed articles, blogs, podcasts, and he's even co-written 2 books. Teaching the iGeneration, and Personalized Learning Through Voice and Choice. Adam is a former educator and is currently supporting school districts and companies across the US and Canada with generative AI initiatives. Welcome to the show, Adam.

Adam Garry:

Thanks, Rich. Good to see you, buddy.

Rich Henderson:

Good to see you too. And Adam, I've been following your work for a while now, and what I've seen most recently is that you're working with a lot of school districts around their AI implementations, their thought process around AI overall. And I guess where I would like to start the conversation today is, where would you say most districts are with AI? Are they still policy setting? Are they experimenting? Are they implementing? Where do you feel like districts are currently in their AI journey?

Adam Gary:

I think they're sort of tangled up in all of that. To be honest with you. Like a lot of them jumped out and said like, hey, we just need a course on AI literacy. And they built something, threw it in the LMS. And then, you know, a lot of conversation was kicked up about policy. And quite honestly, policy isn't really what we need. We need like guiding principles, right? And the policies can come from the guiding principles if that's what you need. And then, you know, some of them have really kind of taken it that next step and said like, how do we sort of think about this and, you know, a very systemic way.

Adam Gary:

So what do we build to sort of roll this out to make sure it's safe and secure? And then once we have all those documents in place, how do we create an AI literacy plan and then ensure that everyone is on a path and start educating folks? And I would say there's a very small percentage that are sort of in that bucket for sure. And Serpi has done a bunch of research on like these early adopters and it still falls sort of in that I would say 10% of district sort of area.

Rich Henderson:

You may know this, but one of my favorite books is Crossing the Chasm, and we talk about those early tech forward, early adopters, and then you've got this chasm that you have to leap over to get the rest of your adopters and, and your late adopters through that process. And I do think it is important to set those frameworks and policies up front. With district administrators, with those who are responsible for the safety of this technology going out into the hands of students. So what do you feel, I guess, is the kind of the keys to understanding those policies and processes that should be in place?

Adam Gary:

So key number one is just even understanding what this technology is and what it isn't, right? Like, if you're going to be building the tools and/or leading the innovation, but you don't really understand how this works and what are the risks and possibilities with it, it's really hard to sort of even create the documentation or update your acceptable use policy. So that's sort of the first layer of this. The second layer is someone has to sort of also be keeping you updated because this is an emerging technology. So the example I would use is like a year ago, as I was helping folks rethink their acceptable use policies, Everything we put in those documents made sense. And then along came the Meta glasses, which were not there a year ago, and now we've sold 7 million copies of these. And now we got to go back to those documents and relook at the acceptable use policies through that lens, right? And so I think it's a matter of A, being educated, B, staying on top of this as an emerging technology, and then C, sort of looking at this as whole system, not just about teaching and learning. Not just about like, how do we get this in the hands of kids so quickly just because, you know, it's showing up in society so quickly, but really thinking systemically about this and rolling it out in a way that makes sense. So before you put it in the hands of kids in your district, you should pilot, you know, before you decide what this looks like, you shouldn't just say everyone needs Gemini.

Adam Gary:

You should say like, maybe kids in 6th through 8th have NotebookLM first because it's much more controllable by the teacher. Before we give a full chatbot to someone, right? Like really taking the time to dissect this, figure out what's best for your district, but also making sure that everyone's on a path for AI literacy because this will impact everyone in the system.

Rich Henderson:

I want to come back to those use cases in a second, but I thought I'd get your take on the fly because I just saw this morning Apple announced that they're actually going to make pendants. They're going to put cameras in their AirPods, right? And we know students walk around with these in their ears all the time. So as we think about adapting to this frequently changing technology. I think that is a really key part of adopting AI. This is not a one-time process that we're going to set and forget this. This has to be revisited constantly as part of a policy guidance document.

Adam Gary:

So there's a couple ways to look at that, right? One is this idea the technology is emerging so quickly, and there's one thing of like Hey, how do we integrate this into our school system? Because it makes sense and it's going to help us with teaching and learning or operations, right? There's the second side of this, which is like, how do we play defense? Because these companies are creating things so quickly and they're actually harmful and could actually, you know, create situations that are not good for learners or for the adults in the building. You know, the glasses that I mentioned earlier, the pendants that you're talking about, but even things like, in ChatGPT, right? Where they decided, hey, we're gonna create a chat feature so that Rich, you and I could actually start a chat together based on something we were prompting. And we can go back and forth and then say, hey chat, what do you think? Well, that's amazing for you and I to be able to organize something together, like a podcast. It's terrible in a school district that works very hard to shut down all these chat features for kids because of cyberbullying and, and other things like giving out test answers and stuff. And they're not consulting with a school district to say, hey, should we have this functionality? They're just adding it, right? And then if the school district isn't blocking ChatGPT, then it just shows up. So we sort of have this bifurcation of things that we have to think about in a school district, which is how do we adopt things that make sense and how do we protect for those things that are showing up in society that could be harmful?

Rich Henderson:

And I love what you said about not letting the tool be the thing that sets the outcome that schools are looking for. We should start with the use cases that are needed and the gaps and the struggles that we have in K-12, whether it be in the classroom level, at the individual student intervention level, or, you know, my background is in special education and I think about that, that use case. One of the areas I know you've been thinking about is high-quality instructional materials. And I kind of want to talk about that for a few minutes. How are curriculum leaders thinking about AI and how are they using it in their curriculum tools?

Adam Gary:

Yeah, I've been fortunate actually to partner with some really innovative curriculum leaders around the country on this, and I sort of see it falling into 3 buckets right now. But I'll tell you, it's like, as I do the work, like more and more of these sort of different use cases pile up and it's getting to be a lot of fun. But like the first 3 out of the gate that we've really been seeing is one is like, well, we already have curriculum, right? Like we've already been using some sort of template for backwards design or, you know, something they've built and they've built curriculum, but they don't really have feedback on it. Someone sits in a room or over the summer a teacher builds something and then they just share it out. Maybe AI with a rubric that's clearly defined can give you feedback and help you add some new content to your curriculum based on that feedback, right? So that's sort of use case 1. Use case 2 gets into this idea of just building complete units, right? So now we've built the template and some districts have it, some don't. Like, we'll start from scratch and build our template. And then what we do is we ground it in the standards, we ground it in all of the documents they've been creating for all of these years, right? About high-quality instruction.

Adam Gary:

So we make sure that this, the chatbot understands exactly what we mean by unit design. And then we give it our scope and sequence and we say, hey, build these units for us. And then we bring in the humans, right? Over the summer and/or throughout the school year and say, now as a teacher, someone who would be teaching this unit, tell us what we can do better. What makes sense? What doesn't? Where do you need support? So that's sort of two. And the place where I see that sort of showing up a lot too, Rich, has been in the CTE area, which I know you have a lot of experience with as well. There's courses that are out there and states might give them, you know, an overarching description of the course, but no objectives, no scope and sequence, and no units, right? And I've been working with, actually with Bakhtia in Virginia, and you know, we built out a whole process to get from nothing to whole units within 10, 15 minutes for a course. So I think there's some real good use cases in that.. And then the last one is really around supplemental material.

Adam Gary:

We have lots of supplemental resources, but they actually sit outside of the curriculum resources we give teachers many times. So we can use the alignment to standards and build the supplemental resources directly into our units. And like I was just prompting for a district I'm working with this week, and you know, when they don't give us the performance task the way we want it, we could say, build out this entire performance task. Or like this one, it gave me the vocabulary words, but it didn't give me the definitions. Well, you know, build out all the definitions or build out all the formative assessments. Here's our formative assessment bank of questions, right? We can really get to some of that supplemental pretty quickly. And all of this is with the caveat of then we put the humans to look at it and, you know, make sure that there's a human in the loop.

Rich Henderson:

Well, that is a lot of really interesting applications, and it made me think of several areas that I wanted to dig deeper, but I guess the last one that I thought of was my wife started as a middle school librarian, is now a high school librarian, and about 6 or 7 years ago, this was before we were all using or had access to generative AI, she was asked to teach a digital citizenship and essentially a business essentials class in her her library. And I remember sitting down with her and creating that curriculum from scratch for her to teach in her library in center of rural North Carolina. And I remember how difficult that was, really not having even that framework at the time to teach some of this coursework and how valuable now, as I look back, that would be to have really AI able to go structure that coursework, to set the pacing, to look at different activities, and, and really to help funnel ideas and structure into that. It was a lot of fun, but I think we could have been a lot more effective and saved her a lot of late night hours from having to worry about kind of crafting brand new lesson for that next day. And as I think about what Panorama has been working on, where I've seen a few districts really do compelling work is really enhancing their existing instruction materials with individualized or personalized instruction, or like you said, supplemental materials or even scaffolding materials for students based on those students' capabilities, assessments, and skills.

Adam Gary:

Yeah. I mean, that's sort of the holy grail, right? Like what we're building, right, is this idea of How do we just build curriculum in a process? But what you're talking about is how do you then make that curriculum come to life, right? And sitting inside a system like Panorama, now you have access to a whole new set of functionality that then allows you to tie that back to all the other resources that you have so that kids can personalize on their own, parents can, teachers can, right? Like it creates a very different flow of how education happens, which is where we want to get to. The other thing you said that sort of made me think of something else is, you know, that process you went through with your wife was long and hard, and you definitely could have found efficiencies with AI. What I sort of struggle with in working with school districts is like over time, let's say we shave 60% of the time off of building. How do we use that time though, to really go deeper into understanding what kids learn and how they learn, right? Because I don't want it to be like, well, you know, the computer, the AI was able to build the unit so quickly and the teacher really doesn't understand what's in the unit, so they can't go deeper into the questions with the kids. So it's sort of like, what is that balance of what do we do with that efficiency and time? Because there's always these conversations you hear where AI is going to save you time. I'm like, teachers don't have time as it is. Instructional leaders don't have time as it is.

Adam Gary:

Is, right? Like, it's not going to save you time as much as what is it going to do is it's going to change the way that time is used because they show up at the building and they have this amount of time that they're there, right? Like, how do we use the time more efficiently, I think, is a better way to look at it and less about just saving people time.

Rich Henderson:

That made me think about the fact that in the curriculum world, we have so much context to work with. We have all of these standards that have been designed that are specific to the region. We have pacing guidelines, we have rubrics, we have content, decades, hundreds of years of content that we have to apply to curriculum. And where I've been thinking a lot with AI is what makes AI outputs great is really great data. Surrounded with context. And then of course, like the guardrails to make the outputs safe in the, the context, exactly what you need. But that does seem like we are getting to a place now where we have data integrations and we have context that we can supply to these AI models so that the large language model outputs are actually tailored to what we need and they're specific.

Adam Gary:

To that student, to that classroom. Again, like that's where like these units are at a much higher level, right? Then we start getting down to the instructional model. That's where at that classroom level, you're going to see tools like what you're doing at Panorama actually start to really change the flow of how information happens and really get us into competency-based models that we've been looking for for years, right? I think part of the struggle also is like, everything you said and portrait of a graduate or profile of a graduate, right? And all these things that we've been building in school districts for so many years that are really good, but have a hard time sort of making their way into the instructional process because it's just too much. So how do we use AI to ground these systems? I keep coming back to that, right? Because that grounding is so important. We don't want an LLM just giving us the output. We need to ground it in what we know is good as the educator. And that's where I see part of the opportunity here as we move forward.

Rich Henderson:

I want to talk a little bit more about that grounding element, and maybe there's a district that you can think of who has started down that path or has done a great job of it. But also, I'd love for you to talk about what happens if they skip that grounding step. What does it look like?

Adam Gary:

When you skip the grounding step, you get a lot of hallucinations. You get lesson plans that are based off of what we know is on the web. And what do we know about lesson plans on the web? Most of them are terrible, right? Like you're not getting really good outputs. And even the structure, like you could go in one day and say, I need a lesson plan on this and it'll give you this structure. You go in the next day, it'll give you a completely different structure, right? So you're missing out on, you know, making sure that teacher is using a template that actually is research-based and what you know is good for teaching and learning. The example I'd give you on the grounding stuff is in Loudoun County, I've been doing this work and we had an elementary group in there. And again, my job is to facilitate the understanding of, you know, how the tools can be used. But since I have a background in curriculum design, I can work very closely with these folks.

Adam Gary:

The elementary team decided they were going to tackle this idea of interdisciplinary units, but here's the kicker. It's like, They already have a Mifflin's literacy product. That's what they use as the core of their literacy instruction. What was missing was any ties back to science and social studies throughout that day. So that's what they tackled, and they literally grounded it in their instructional model, and then they built a template for what interdisciplinary units would look like, and they kept it simple. They didn't say we need science and social studies. They said like a science or a social studies tie. So that if they were reading these things in the morning, in the afternoon, they were working on these things in science or social studies.

Adam Gary:

And I thought it was pretty aggressive, to be honest with you. I was like, I wonder what's going to happen. Do the models even know what the HMH model is? Blah, blah, blah. And it turns out they did know what was in that HMH material. And they've built these amazing resources that are now interdisciplinary units. And all they had to do was ground it in what they wanted for an interdisciplinary unit and their literacy.

Rich Henderson:

Model. And I think we're getting to a point with prompt engineering and interacting with AIs that we're understanding that we need to set the framing. We need to give information into the tool to set the framing, but we also need to provide how do you judge whether this is good or not? So we have to provide kind of a rubric to the AI in a way to like know when it's met the goal that we're looking for.

Adam Gary:

I agree. That's why, like I said on that, like first tier of like sort of just getting feedback, like uploading a unit and saying, hey, give me feedback is a terrible way to do this. When you go through that process though with the curriculum team, if they don't already have a rubric, the kinds of conversations that you have about teaching and learning are amazing because now you're defining essentially what good instruction looks like. Because if you don't and you give this rubric to the AI, it's just going to give you feedback on what you gave it. So, It is, it's like a super important process to go through as a curriculum team, whether you're going to use AI or not.

Rich Henderson:

We have this AI writing coach that I need to show you at some point called Class Companion. And in it though, one of the most important parts of creating that assignment is what is the rubric that the AI is going to score that student on? And we've been encouraging our users and our teachers to spend much more time kind of thinking about that rubric. And how it's going to score their response versus kind of the assignment question itself.

Adam Gary:

And sometimes the rubric is more important than that question format. Yeah. And I mean, again, it's about feedback for kids, right? Yeah. The number actually isn't even as important because maybe that's left up to the human at the end, but the actual language of the rubric so that it gives proper feedback based on, you know, a student's voice, right? And within that writing. Becomes the most important element of what you're doing.

Rich Henderson:

I did want to follow up on something we talked about earlier, which was about these CTE pathways, for example, because part of it is the content and the curriculum that needs to support this structure. But also what we're finding is that there are so many more graduation pathways available for CTE. There are seals of distinction, there are these competency requirements. There are these college and career courses that are now being given credit in high schools. There's so much complication there in the, the graduation pathways aspect that we are working at the state level in a few states now, helping to define those pathways and leveraging AI to essentially individualize graduation planning for students with all of these new pathways, but also what's available to them regionally in terms of scholarships or job opportunities or community colleges per se. Have you seen more of that in terms of like CTE pathways and how schools are dealing with that?

Adam Gary:

Yeah, I mean, like, it's a big conversation everywhere I go, right? Because I think schools are starting to wake up to this idea that we can't just say college, career, or life and not really prepare them for any of those paths, right? Like, meaning not that we're not preparing them, meaning like that they do have a choice in what they decide. And sometimes it's not to go on to college, it's that they want to go start a career and they want to work on a certification while they're in a school district. I think there's two ways to look at that, right? Like what is the tracking, the opportunity, the helping kids through that process? That also gets into counseling, right? Not just, you know, the CTE director's job. So what's that crossover? The other way to look at it is like, how from the state do we make it easier for school districts to adapt and to build courses so that kids do have opportunities to learn about these things? Because if it takes 3 years to get a new course in AI and machine learning, you know, you're sort of missing the boat on some of this stuff, right? So, you know, how do we enable school districts to adapt much more quicker to create these courses?

Rich Henderson:

As we begin to wrap up the conversation, Adam, I would love to hear, we're gonna have superintendents and, and other leaders listen to this podcast.

Adam Gary:

How should they think differently about AI than individual classroom teachers? You know, they are the system leader. That's at the end of the day, they have to be thinking about this from an entire system and not just teaching and learning, but I'll sort of break that down for them though. On the teaching and learning side, they can look at the 3 ways or some other ways that we've described in the podcast, but they really do have to think about how they're going to leverage it, not just the use in the classroom, but how do we build and create things for teachers and for students so that they can ensure that there's high-quality instruction going on in that classroom. And then on the opposite side of that, it's sort of the operations piece. And how are we augmenting work on the operations side? Because we all know that there's not enough people for the work that these folks have to do. So what processes and things are we doing to ensure that they're educated and then can augment the processes? Because they have a certain domain expertise that in many districts, they're the only one that has that domain expertise. So we can't build stuff for them. We have to build stuff with them.

Adam Gary:

And that becomes an important part. And then Lastly, I would say like the summer is a time where we bring people together, especially on the curriculum side to build. And this is a huge opportunity to really bring people together in a place where we have much more for them to look at and give feedback to instead of having to start from scratch if we leverage these tools appropriately.

Rich Henderson:

I love that notion. I think about this a lot that artificial intelligence should empower better human decision-making. And when we can use the time to plan, like many of the ways that you were describing today, to use AI more effectively, we actually free up our educators to use our, our better human intelligence to interact with each other and to plan together. Well, Adam, thank you for walking us through that today, and I really appreciate the discussion on how curriculum teams are wrestling with AI. Assessing what exists today, building where there are gaps, and, and really connecting these supplemental resources to scaffold learning for their students. Thanks again for joining us and sharing the work you're seeing across districts.

Adam Gary:

Yeah, you're welcome. Thanks for having me on.

Rich Henderson:

I appreciate it. If you are listening and leading this work in your district, I hope this conversation gave you a clear lens on how to move from experimentation to intentional strategy. If this episode was helpful, share it with your colleague who is thinking through AI and curriculum alignment. You can find show notes and ways to connect with Adam in the episode description. We will see you next time.