Transformative Principal

Thomas Thompson, founder of Eduaide.ai and a Maryland teacher, shows how AI for teachers can speed up lesson planning while leaving the creative work to the teacher.

Show Notes

Thomas Thompson, founder and CEO of Eduaide.ai and a Maryland teacher, explains how AI for teachers can cut planning time without replacing teacher judgment. He and Jethro Jones cover teacher turnover, the source it, shape it, share it workflow, and where the technology falls short.

Thompson says a teacher spends about six and a half of eight hours in front of students, leaving two hours for meetings, email, lunch and planning. His fix is a lesson seed, a range of activities and assessment questions, that the teacher then refines the way a sculptor shapes marble.

What you'll learn

  • Use AI for rote planning tasks like finding resources and translating, then spend the saved time personalizing instruction for your students.
  • Treat AI output like a block of marble: start from a lesson seed and refine it instead of accepting a finished plan.
  • Give students six AI-written essays graded D, C, F and A, and ask them how to make each one better.
  • Check AI output for subtle bias and hallucination, and write prompts with examples and parameters instead of one-shot requests.
  • Skip AI for tasks you can do faster yourself, like Jones writing podcast show notes after an interview he conducted.

Chapters

  • 0:00 Why Thomas Thompson started Eduaide
  • 2:49 Teacher attrition and the planning time problem
  • 5:39 What AI should and should not do
  • 10:11 Lesson seeds and the sculptor's block of marble
  • 13:58 Students evaluate AI essays instead of writing
  • 15:21 AI tutors, chess bots and their limits
  • 23:14 Bias, hallucination and prompting with best practices
  • 28:43 How AI lowers the cost of trying new methods

Times are for the episode without ads. Your player may be off by the length of any ads.

Questions this episode answers

How can AI help reduce teacher burnout?
Thompson says teachers get about two hours a day outside instruction for meetings, email and planning, so work goes home. He argues large language models can condense rote tasks like finding resources and translating, freeing time for creative planning.

Can AI write a good lesson plan?
Thompson says generative content is like a block of marble, not a finished product. Eduaide offers a lesson seed, a range of activities and questions, so the teacher's job shifts from creation to refinement. He warns that bias and hallucinations remain limits.

How can students use AI without it becoming cheating?
Thompson suggests the teacher generate six essays that deserve grades from D to A and have students work out how to improve them, or have students read an AI article and find flaws in its argument. Students evaluate quality instead of writing.

About Thomas Thompson

Thomas Thompson is founder and CEO of Eduaide.ai, an AI tool for teachers, and a Maryland teacher with a master's in educational technology from Johns Hopkins.

Links and resources

Working on this at your school? Get a free preview of a 10-week leadership pathway built from episodes like this one at transformativeprincipal.com/start.


Thanks to our Sponsor Broadcast Made Easy who help you start a news broadcast at your school by having everything you need in one easy-to-download place. Check it out today and get your kids creating. 

We’re proud to be sponsored by IXL. 

IXL’s comprehensive teaching and learning platform for math, language arts, science, and social studies is accelerating achievement in 95 of the top 100 U.S. school districts. Loved by teachers and backed by independent research from Johns Hopkins University, IXL can help you do the following and more:
  • Simplify and streamline technology
  • Save teachers’ time
  • Reliably meet Tier 1 standards
  • Improve student performance on state assessments

🚀 Ready to see why leading districts trust IXL for their educational needs? Visit IXL.com/BE today to learn more about how IXL can elevate your school or district.

Creators and Guests

Host
Jethro Jones
SA
Guest
Summer of AI
The Summer of AI Series included folks from a range of perspectives talking about AI.

What is Transformative Principal?

Jethro Jones interviews instructional leaders from around the nation to learn and teach what it takes to become a transformative principal. Episodes address topics like Response to Intervention (RTI), Multi-Tiered System of Support (MTSS), Positive Behavior Interventions and Supports (PBIS), the latest advances in educational research, standards-based grading, and interviews with industry leaders like Bill Daggett, Rick Wormeli, Todd Whitaker and even people outside the education like Seth Godin, JeVon McCormick, Liz Wiseman, and more.

If you've been thinking about news broadcasts but don't know where to start and don't want to have that lift, use BroadcastMadeEasy.com to make it super easy for yourself. I am really excited about our sponsor, Broadcast Made Easy. As a principal, I created a daily news show at all of my schools, but you know what? It took a ton of work. Broadcast Made Easy makes it super simple for you to get started.

They provide templates, scripts, job descriptions, applications, the whole nine yards. You buy it once, no subscription, and it makes getting started so much easier. Another thing they have is specific videos for every holiday you could possibly imagine so that as part of your news broadcast, you can share what holidays are coming up, less than a minute, and very simple, straightforward, just the facts that makes it really easy to start conversations about those holidays.

So go check it out at BroadcastMadeEasy.com and let them know that you heard about it on Transformative Principle. That's BroadcastMadeEasy.com. This B Podcast Network show is presented by IXL. IXL's all-inclusive online teaching and learning platform simplifies edtech needs and accelerates achievement in 95 of the top 100 U.S. school districts. IXL delivers personalized learning across a comprehensive pre-K-12 curriculum, including math, language arts, science, and social studies, and helps you assess student performance through actionable, real-time insights at every level of your school or district.

This one solution performs work that typically requires dozens of tools. Want to find out why so many leading districts trust IXL? Visit IXL.com slash B-E. That's IXL.com slash B-E. Okay. Welcome to Transformative Principle. I am so excited to have Thomas Thompson on the program today. He is the founder and CEO of EduAid.ai, which boasts over 100,000 users. EduAid is an AI tool to help eliminate teacher burnout and make high-quality teaching resources universally accessible.

Thomas has a master's in educational technology from John Hopkins School of Education and is still a teacher in the Maryland area. So running a business and being a teacher still definitely gives him insight. Thomas, welcome to Transformative Principle. Appreciate having you here. Thank you for having me on. I'm excited to sit down and talk. This is going to be, if you've been listening to Transformative Principle for any amount of time, you know that I have a love-hate relationship with ed tech.

We have the ability to personalize learning for every single one of our students. And yet, so many of our ed tech tools fall short. We need our technology to do more for us. That's why it's so important for me to know that IXL provides true personalized learning across the entire pre-K-12 curriculum. And that it's proven to benefit all student populations, including English language learners and students in special ed programs.

As a principal, I've used this in my school. As a parent, I've had my children use it as well. And let me tell you, this is a tool that definitely helps students learn and practice better. IXL is research proven to accelerate achievement. Studies across 45 states show that IXL schools outperform non-IXL schools on state assessments. And independent research from Johns Hopkins University verifies that IXL meets ESSA Tier 1 standards.

With those results combined with IXL's teacher-friendly reputation, what more could you ask for? Now, you also know that I don't care so much about test scores, but I know that they are legislatively convenient and something that we have to deal with and manage on a day-to-day basis. If you can implement something that is easy and effective, why wouldn't you do it? If you have a goal to increase achievement for all students, make sure to find out what IXL can do for you.

Visit IXL.com slash B for a demo. That's IXL.com slash B E. A great conversation. But why don't we start by you telling us about how and why you started EduAid? Because you have an interesting career path and you've done some interesting things. And I think that this would be an interesting insight to get started with. So I'm calling you right now from Maryland, but I'm not originally from Maryland.

I'm from Western Pennsylvania, North of Pittsburgh. My co-founder, Thomas Hummel, is as well. But he and I never knew each other, even though we grew up about 15 minutes away. We met each other in Maryland. We were both teaching middle school right across the hall from one another. And he and I shared many, many conversations, very similar to probably the one we're going to have today just about the institution that we work in, just the nature of education being this public good and foundational human right, and a number of the concerns that get in the way of it being done well.

And just a lot of just maybe just banging our head against the chalkboard kind of conversations. And out of those came a few ideas, but nothing really came of fruition. He moved to another school. I moved to another school closer to home, saved myself an hour commute, undertook my graduate work. And then we reconnected about in March of this year with the idea that perhaps large language models can offer an opportunity to solve some of the problems that he and I discussed.

Not all. We don't think the technology is a panacea. It's not going to solve all of our educational challenges. However, we saw it as a way to solve a few key challenges pretty well. And really, we started working on that from that realization on. Yeah. So what are some of the problems that you're setting out to solve? The first one, of course, is teacher attrition. I don't know how you build an institution that values excellence if you cannot retain talent long term.

The percentage of teachers that turn over within the first three years is unconscionable, and we could do better than we do. So why do we think large language models can help solve that? Well, we view large language models as a way to increase efficiency in the teachers during their planning time, right? Let's say I have an eight-hour workday. Six and a half of those hours are purely instructional on average.

That means I'm in front of the classroom. I don't really have time for anything else. So I'm interfacing with my students. The other two hours of my day are tied up for everything else, right? IEP meetings, parent meetings, faculty meetings, staff meetings, team meetings, department meetings, so on and so forth. Planning. Lunch. Lunch, fielding emails. I have a terrible addiction to coffee, you know, getting a cup of coffee during that planning as well.

And consequently going to the bathroom a lot. Yes, yes, yes. I was up twice using the restroom. Thank the Lord that I have a co-teacher. That makes my life much easier instead of having to sneak it in between classes. But that amount of time is not enough to do effective, deep work. Instructional planning isn't a rote task. You just come in and do the routine. It's a task that requires a level of creativity and a level of ongoing professional development while you're doing it.

You might want to look into a technique or a method you want to try out and you want to do some exploration. Well, you end up taking a lot of that work home with you because you have to do also grading, feedback, fielding emails, and the like. So now you have this issue around the work-life balance and what sacrifices are you making to do the work you enjoy to do?

And many teachers feel as though that isn't, that's not a great state of affairs, having to take a majority of the work home because you don't have a time during the day. And that leads to burnout, that leads to turnover, that leads to people moving on to different careers. And we said, okay, expanding the amount of time you have to plan, that's a big ask and a big pull and it's going to involve contract negotiations and unions.

And just years and years and years of just being bogged down in that kind of process. So what if we can condense the amount of time a teacher would spend on any given task so that they can unlock some time for themselves to engage in these creative tasks and the exploration of lesson planning? And that's where we saw the power of large language models to increase teacher efficiency and effectiveness.

And then hopefully in doing so unlock time so that we can increase teacher excellence over time. Okay. So interesting that you're saying that teachers need time to do deep work, to be successful with planning. And, and yet you're creating a tool that does the planning for them. So talk about that aspect because some people may hear that and say, well, if it takes deep work, then how can a large language model just take care of all that for you?

And tell us what you mean when you say condense some time so that they have time to spend in those other creative pursuits. Yeah. So we do not envision the AI doing everything. It's very, it's limited in what we actually want it to do and maybe more limited than we currently have it on our application. What we mean is, let's say you have to do a plan a lesson on British colonial policies in America.

And all you really have is maybe that keyword to go on or some objectives from your, from your state framework. And you spend a lot of time searching for resources that you might want to use. And you're jumping between all of the different tabs, right? Your school bought a resource package. You have the textbook package. You're reading through all of these documents. You're pulling them together. And then you have maybe some students who are English language learners and some students with IEP.

So you have to differentiate instruction. You have to translate instruction. Now you're jumping between your translate tab and your other app to scale the Lexile level. We said, okay, let's cut out the friction points. Let's bring all of these tools into one place. Use the power of generative content to replace these very restrictive, static, copywritten resources that you might get from a textbook company. So by bringing these resources under one roof and making them all really small and modular, we're hoping that the AI doesn't do everything, but rather it does a few things the teacher would otherwise spend hours, say, searching for.

And then that will give them the time to use a suite of personalization tools that we offer on the app to shape the instruction to be exactly what they need. I mean, on our landing page, we have a, just a description of our workflow, which is three steps, source it, shape it, share it, right? Source it being the only the first step. That's where you're actually generating the content.

Shape it is where we hope the teachers will spend the most amount of time where they're differentiating instruction, personalizing instruction, extracting keywords, making glossaries, looking at new methods and techniques to explore and conveying this content with their students and only then is it ready to be shared. So we free up the tech, free up the time on the rote tasks, like finding the resources and perhaps maybe translating and things like that.

And then open up the amount of time we have for personalization and differentiation where the teachers can do that creative thinking about the unique needs of the diverse learners in their room and then, you know, save that initial rote time. They have the time to explore those opportunities. This is something that I think a lot of people who are hearing about artificial intelligence and teachers using it aren't understanding about what it is that a lot of people like you, like me, are trying to, to accomplish with these large language models, which is the goal is not to have it do all the work for you.

The goal is if I understand you correctly, correct me if I'm wrong, the goal is to make it so that the time and energy spent on things that don't need the personal touch of the teacher that don't need the human element of the teacher are created much faster, quicker, and easier for the teacher. A good example is for like assessment of questions, right? Creating assessment questions for me personally, as a teacher has always been a terrible, awful, horrid thing that I just despise.

And being able to ask a large language model to come up with questions, those at the very beginning are going to be better than what I would come up with, because that is just a difficult task for me to do, especially if it needs to be multiple choice and coming up with wrong answers that show that the student really understands it. It's just, it's just a big ask for me. Is that, is that what you're talking about when you, when you're talking about what the AI tools can do?

That, and also as a kind of a creative collaborator, it gives you a starting place, right? I might ask, we offer on our website, it's our most used resource, a lesson seed, which is different than a lesson plan. It's not a prescriptive document as much as it is, you could do this range of activities to meet this objective, or you could do, ask this range of formative assessment questions. And then as a teacher, your job is no longer pure creation as much as it is refinement. You take what is there and hone it, right? We, I kept coming back to this, this sculptor metaphor when we were first starting Eduade, which was like, we should look at generative content, like a sculptor might look like a block of marble or something, right? What exactly can you do to this to make it that work of art or that, that thing worth some value? It's the marble might be a fine quality, but it's not the

finished product until the artist puts their hand to it. Just like the generative content itself may be of quite fine quality or, or not all the time. I mean, there are limitations with biases and hallucinations and the like, but what can the teacher add to it that will create that, that thing of value. So let, let's talk a little bit about the, the decisions that go into it. And if, if we're like going back to those multiple choice questions, like I intentionally chose that because multiple choice questions are like the worst questions we can ask. We only ask them because they are easy for us to grade and see a response to. Whereas students being able to express what they've learned in their own words is so much more valuable because you can really see what it is that they've learned. Uh, but it's just really hard to grade. And so that's one of the things that AI can potentially help with as well, that it

doesn't just need to create an, an recreate another bad system. Uh, there can be some change there. And with this approach, you have opportunities, you know, a couple of weeks ago, I had a Kalen Sharp from school AI.com on, and he talked about how they're creating these spaces where kids can go in and have conversations with the AI and the teacher then can get a different perspective about what the kid understands or doesn't understand that is more conversational and allows the AI to assess whether or not the student is getting what it is that the teacher is trying to teach them, but do it in a way that is not like answer ABC or D. And so let's talk about some of the ways that we can change how we teach with the advent of these types of tools. What are some of the things that you're seeing or that you're trying to, as you design your tool to kind of force some of these changes to happen?

Well, first, what you just said was quite interesting, right? That is a best practice in education, elaborative interrogation, where you have a student explain and elaborate exactly what it is they mean by the statements that they make. Using a chatbot to do that, to elicit those kind of elaborative explanations from students would definitely be a way that we could better align instruction with best practices in education as discerned and ascertained from, say, the cognitive science research base or learning sciences more broadly, which is where I would prefer to pull a lot of our methods from. I feel like we have in education kind of a, I don't want to say a habit, but maybe there is, I noticed maybe in more than other professions, a habit to pull away from a research-based approach. Even though we say we're data-driven, right, it ends up feeling like we collect a lot of data in education and there's not a lot of time to act on it. To your question about what this

technology could do, there's some interesting implications like maybe a teacher generates six essays from a language model, tells the students these are Ds, this is a C, this is an F, this is an A. How could you make them better? Now you have students engaging not in essay writing, but the consideration of what is it that makes a quality essay, a quality argument, right? Now you have the student doing the work of the teacher in ways that are additive for the student. Or you could perhaps have an assignment where you say, prompt the AI to do X, Y, and Z, or create an article about a certain topic or a certain theme. And then you have the student read the article and identify whether or not the artificial intelligence actually makes that claim or makes that argument or what flaws are in that argument. It takes the process out of the student having to go about and write, say, a full essay

where we're now so worried they could use AI to cheat. But consider what it is that makes a good essay, what it is that makes a good argument by interfacing with the AI, looking at examples of what it generates and finding evidence inside the generations, doing a lot of more fact-checking kind of informational literacy that I don't think was before possible. Now, of course, there's the use cases that we're already seeing development for, which are artificially intelligent tutors.

And there's a lot of great systems in the works that are coming out. Latest announcement from Khan Academy, their AI tutor. I mean, we're going to see AI tutors out of the wazoo. In fact, we were pitching around an idea. Alexander the Great called Aristotle his pocket philosopher. And I always thought it would be great now if you trained a language model on the complete corpus of work on Aristotle or some ancient Greek philosopher who we have a lot of the surviving work in the public domain. And then you have a pocket philosopher where you can, say, select Marcus Aurelius and ask him all of your questions.

That'd be quite fun. That'd be an interesting application that I would certainly use. But the use of these intelligent tutors who are trained on the data of different individuals before we were talking about the philosopher with Wittgenstein, it'd be interesting to train a model on him and have a conversation with AI Wittgenstein. And that already happens. I mean, great podcast conversation with conversations with Tyler, with Tyler Cowen. He interviewed a chat GPT plugin with Jonathan Swift that was quite interesting and a lot of fun. So those avenues for these intelligent tutor models are going to be interesting. And we've already seen applications of it. I mean, in chess, for example, a lot of people train against chess bots because they're fantastic trainers that never get tired or exhausted and they'll always make the right move if you want them to. And that's, it's increased the play. I mean, more people play chess now than did a hundred years ago or even 50 years ago.

There's a, it gets me excited because these, these intelligent tutor type ideas are the ones that I think will bear out and have borne out a lot of, a lot of value in different fields. Again, like chess. So in that specific example, the value becomes really powerful when you think about a specific skill set that you're trying to teach where they're like chess, for example, where there's a lot of information about what makes a good or great chess player and being able to scale that up or down in a way that is, that's really powerful to help the person learn. So could you have a, we'll stay away from math because they're still so bad at math, but could you have a, an English writer as something that would help you? Could you have a scientist who could help you learn specific scientific principles that are, are that worthwhile? Could you have a coach for a sport like football, basketball,

tennis, whatever it is to constantly give you feedback on every single play, every single thing that you do, those applications become really fascinating and very different from what we're currently doing and have the capacity to do right now. I mean, maybe the drawback that I pull from, from my own argument about the, the chess example is that chess is a closed system, 64 squares. There's only so many permutations and of, of chess that said that it's a huge number, right? Um, it expands rapidly the deeper into the game you get. However, I mean, then there's the question like, is, is, is language a closed system? I mean, chat GP is, chat GPT is demonstrating that it certainly maybe feels like it is. And I'm curious what these models will look like in five years, but other things like different kinds of games. I, I feel like you probably could see interesting AI tutors, but then there's a lot of the, the disembodied human movement type things, like how

valuable could it be in telling you how to play basketball? Um, it's a language model. So there is this disconnect between our world as language and our world and being as, as it actually is for us physically. Yeah. And I, and I think there's also this different perspective as well that I'm, I've had this idea in my mind for a while and I haven't been able to articulate it well. So forgive me as I stumble through it, but there's this idea of proficiency that we strive so much for in education where, you know, we pass off specific standards or achieve certain goals in what we're trying to learn. And while that is all fine and dandy, just because you know, something doesn't mean that you can transfer to actually doing it. Right. And so I can understand the mechanics of how to shoot a basketball and, and even be able to tell myself, apply a certain amount of force to this shot. And I know that I can

make it that distance and angle it this specific way. And I know I can get it towards the basket, but does that mean that I can actually translate that into playing the game? And this extends to many other things as well. You know, I've, I've published two books and I'm writing my third book. I've already written another book that I haven't published yet because there is a mental block saying, this is not done yet. I don't know what to do to finish it. And I'm, uh, and I'm afraid of taking the next step and diving back into it. Um, because I don't have time or whatever other excuses, all these things come into play that I know how to technically write a book, but writing a great book is different.

And this other book that I've already written is a fiction book and I haven't published a fiction book before. And so I'm a little bit nervous about it and whether or not it's as good as I think it is, or if this story is good in my head, but it's not good on paper yet. Those are all these things that I'm trying to wrestle with. What's your thought on that? Well, I mean, for me as, as someone who is engaged in a little creative writing projects in the, in the past, but nothing ever came of them, I always felt like it was very hard to get quality feedback. You were never sure if maybe it's just the person likes you and they're trying to still be your friend. And how do I get this, this hard hitting feedback? Maybe I feel like I can't access it given my circumstances. I think AI would be great

in that regard. Now, as we were getting at earlier, you put in an essay and the feedback was about as long as the essay was and right. Right. Kind of combing through that. But if the feedback is solid and you know, it's pointing out maybe stylistic things, that's interesting. Maybe the more interesting thing is using the AI to not necessarily train you in what makes a great book or a great basketball player, but understands what are the parameters you must be in to have the highest likelihood of success given past examples. So maybe it's not that generative AI will make a particularly good educational planner, but maybe it knows the parameters in which an education, a lesson plan is most likely to succeed. Like maybe there's a set of best practices that have, that have a lot of promise given the many, many meta analyses it could run with a large enough data set. That is,

that's what's interesting for us. And that's what we really tried to do with, with embedding best practices into our generative resources. Like what are the parameters that make say a good multiple choice question, or what are the parameters that make a good team-based activity or a collaborative learning experience or, or inquiry or any kind of learning method. Honestly, for me, it really seems to come down to assessing and student prior knowledge, but that's a different kind of thing altogether.

Here's the truth. School leadership is demanding. It's rewarding, but it can also feel lonely and so overwhelming sometimes. That's why we created linked leaders to make sure great leaders keep leading and good leaders get the support to become great. At linked leaders, you're never on your own. Whether you're building culture, navigating board relations, or tackling special education challenges, you'll find solutions from leaders who've been in your shoes. And the best part, you don't have to wait months for a big conference or spend thousands of dollars hoping to make the right connection.

With linked leaders, you can log in today, get straight answers from experienced mentors, and walk away with a playbook that works right when you need it. Linked leaders is efficient, personalized support that keeps you moving forward. Sign up for free at linkedleaders.com. So let's talk a little bit deeper about that because I think this is really interesting too because when you're creating something, then you're defining how decisions are made, right? And you're embedding best practices into the models and the things that you're doing. What does that actually look like?

What does that look like on a real level? You gave a couple of brief examples, but let's go deeper into one of them, choose which one, and talk about what that looks like. So cooperative learning, for example, we could go down there and say, what makes good cooperative learning? And one, how do you know? And then how do you influence the AI to come back with something that matches that rather than whatever it feels like it thinks should be good?

And they don't think, but you know what I mean? Well, it's at this point that I feel like we really come up on the limitations of the technology because depending on the training sets of the large language model, biases do exist and do permeate the responses. And there are plenty of instances of hallucinations where it'll generate wholly false material. But sometimes those hallucinations and biases are very subtle and maybe not something you would notice on the first go around. Humans have a way of filling in the correct word when they misread or something like that. However, I think there are ways that you can better guarantee success more reliably, right? How could you guarantee success more reliably than say with a one-shot prompt where you ask the AI to do something? So there's some prompt engineering things that you can do. For example, with EduAid, we look at a lot of in-context learning. How is it that we can lead the language model into

being in those parameters of what makes something useful? That might be with examples or with defining what the parameters of a, say, multiple choice question or cooperative learning activity are. I could take, because I was just working on this one last night, I could talk about unit plans a little bit. The broad, broad planning document, four to six weeks of instructional planning. There are some arbitrary things like the four to six weeks of instructional planning. That's based on averages, right? Maybe a unit's longer for some people, maybe it's shorter for others, but it's kind of splitting the difference as best as we can so you have a decent chunk of time.

That's another scary thing, setting the time, right? How long is a lesson plan? For some teachers, it's a block schedule, which is 80 minutes. For some teachers, it's a shortened class, 42. For me, at least, it's 52 minutes. You'll notice that Edu8AI offers a 52-minute lesson plan. I made the thing, all right, it's 52 minutes because I teach 52 minutes. Now, does that also feel like I'm splitting the difference between the two? Yes. Maybe there'd be a better way, and we'll investigate if there's a better way of calculating that time, maybe what the exact average is. But for a unit plan, it's interleaved practice, which was huge for us. Distributed interleaved practice, right? So the to-be-learned material gets brought up again throughout the unit over time. That way, you reinforce the material through this, again, four to six-week period. We know from a lot of, again, the literature base in learning sciences and cognitive sciences as it applies to education that we see, you know, pre-testing is quite an effective technique. Even if there isn't going to be, say, a test,

a test, an assessment to take, there should be an assessment. Quizzes are also usually pretty handy for measuring student learning. So interleaved practice was big. Distributed practice, the checks for prior knowledge at the beginning of each lesson, right? So a teacher can measure whether or not the student is where they need to be to be able to interact with the content at this level. And that usually comes through warm-ups, which then must be included in a lesson. So it's really just this iterative process of given a set of best practices, how do you build something that reflects it? Because you can't always say do X, Y, and Z to the language model and expect it to be aligned to that thing. It's this kind of leading through examples or leading through this parameter-like definition, and then a lot of testing, and then making tweaks when the tests come up in a way that we don't believe reflects what we want,

right? Sorry, that was a huge answer. That was a mega prompt, I suppose. Yeah, no kidding. So that's given me a lot to think about. And what it really requires is, you know, I guess the thing that I'm thinking is the reason why someone should be using a tool like yours is that you've done a lot of that work on setting up the prompt in the background so that the teacher can spend even less time thinking about that idea. So let's say for example, that you have this unit plan and the teacher's needs are different from how you've built and designed this thing, which is naturally going to happen in a lot of situations, right? And so that still requires some level of flexibility, creativity on the part of the teacher. But at the same time, that still is shortening the time spent in preparation, hopefully. And I've done a few things using different tools just to compare how they work and

and seeing that sometimes, like, honestly, I could have just written this faster myself. And so it doesn't always save as much time. And, and there are, there are certain situations that are specific where that has happened. Specifically, one for me is podcast show notes. I can typically write those faster because I did the interview, I talked to the person, I have a better idea of how it really is going. And what I want people to get out of it. And those pieces of intention are something that's another limitation of the AI that it, it doesn't understand what our real intention is. Even if our intention is meet or pass the standard, it still doesn't totally understand how to apply that in a real life situation. What, what are your thoughts on that?

I believe the technology lowers the cost of evolution, meaning it's much cheaper for a teacher now to explore a new method or technique than it ever was before. Maybe you can write a lesson plan much quicker than the AI could, or maybe you don't really write lesson plans. A lot of teachers don't. So you won't just don't, don't use that, make the consumer choice and say, I don't need this, I'm not going to use it. And great, you're playing to your strengths. But when it's time to do the administrative rote tasks where you don't really need that personal touch, and we know that we can reliably get good results, then maybe you want to use the technology. Or when you want to explore new options and you're uncomfortable, right? With say, maybe cooperative, a specific cooperative learning strategy that you always found interesting, but never really implemented in your classroom, maybe like the jigsaw activity, right? Where you have your expert groups and go into

that. But maybe you use an AI now to see what a jigsaw would look like for this content that you know, really well, and you have your lesson plan you are already going to do. But now you can explore this additional option without that significant upfront cost of spending half of your planning, exploring it. Yeah, I think that's a really good example. And a great way to say that that it lowers the cost of evolution because you can try new things with a lower barrier to entry. Exactly.

You don't have to go to a two day Kagan cooperative learning workshop to start bringing some cooperative learning into your classroom that when you don't when you just don't know anything about it. And so you can go to a resource that somebody else has created. And, you know, there's lots of things on teachers pay teachers or blog posts or whatever that's out there that is for their specific things.

But if you need something for you, you can customize it enough that you can that you can get something. What do you speaking of teachers pay teachers? What do you think something like this is going to do to to a site like teachers pay teachers? Well, my my research area in grad school was on open educational resources, which are freely and openly available teaching materials and learning objects.

So I felt as though perhaps it was kind of antithetical to the mission of education to place valuable educational material behind paywalls, even, I guess, particularly steep paywalls. Maybe that comes mostly in the form of research journals. For me, there's huge, huge disparities in which universities and colleges have access to those and which ones do not and where they're located geographically around the world. Same, I would argue, is true for educational resources.

So when I get an email, as I did from a teacher who did a professional development in Pakistan, where they are engaging teachers in the creation of educational resources in their language, because we offer a range of language accessibility options. And then they have this this personalized material. And perhaps they have to spend some time making changes. But they have the starting place that did not exist before. Again, as as someone who does podcasts, I bet you're aware of Dan Carlin, hardcore history, right? He's got this great quote. Yes, he's a wonderful podcaster, just the talent of the guy to do these five hour opuses. But he said that things open up in ways you can't imagine simply because you did something. And that quote has really stuck with me for years now. And I think that's, I think that's true, right? Things are opening up in ways you can't imagine for teachers and for

educators, simply because this technology is now available and openly and freely available. That totally didn't answer your question. To answer your question, EduAid, for example, is going to launch a community in which teachers will be able to openly and freely share the resources that they generate on the platform with other users who could then say, import that to their workspace and say, oh, this was for this entire lesson they have was for a classroom with three English language learners, two students with this particular IEP. And it's like, oh, that's that's my context. I have a very similar situation. And now you can take that. And since it's open, you can remix it and revise it in the app to fit even closer what you were doing in the classroom. So it's, it's installing this level of checks and balances with the technology. You're pairing the artificial intelligence with the collective intelligence of users with the intuitive choices that the individual might make. And we're

hoping that this can create this ecosystem in which the technology becomes more reliable in the, in the longterm, right? Cause the foundational models themselves, chat GPT, llama, et cetera, et cetera. They, they're not particularly useful in their current form for doing the reliable work that we need them or would want them to do in education. It requires some fine tuning, embedding, some prompt design, best practices, and understanding the content area you're looking at. So there's a lot of work to go, but I think it's a good place to start. Yeah. I, I appreciate that perspective. And that's another thing that when we, when we talk about the design of these tools that you are intentionally bringing this open approach to it in, and that influences your, your design decisions about allowing people to share what they're doing, make it public, and then let other people remix from that. And, and that I think is an area where these AI tools are great for getting a first draft. Yes. But then

they're also really great for finding something that is already existing and adding a little wrinkle to it or, or adjusting it to a certain amount so that it's more dialed in for your particular context. And that when you give it that information and say, Hey, I've got, this is great, but I've got a student in my 10th grade class who's reading at a second grade level. I need you to make this accessible for them.

Yeah. That's a, that's a possibility that is, is exceptionally difficult with the textbook that you've got sitting on, sitting on your teacher's desk, you know? Because it's that, it's that static single object that doesn't have this level of, of flexibility. That was another large part of what we were hoping to accomplish was installing this level of granular control that a teacher could take over anything they put in the workspace. Instead of conceiving a generative technology as like a, you click this thing to generate this thing. It might be, you highlight a portion of text, right? Click it. Now there's these cool things you could do to manipulate the text. Like you might manipulate a photo on Photoshop. And that has some dangerous implications, right? That's not always a positive thing. And we can also get into that, but say now as a teacher, I have a text that I love to use, right? It's reliable. I've taught with it for years. There's a lot

of high impact stuff in there, but maybe there's some things you want to add, right? Maybe you want to extract the most high use keywords that students of a particular reading level might have a problem with and provide definitions for that in a glossary. Well, you can do that now. Or maybe you want to scale the reading level up or down. You can do that now. Or you want to insert a couple of those, as you were talking about hate writing multiple choice questions because you have to figure out what a good wrong answer would look like, right? Now you could insert those into the reading. So you make it more of a, more of an active exercise. Like there's things you can do in playing with the text that you already have that wasn't before positive possible. It turns this whole text process into kind of this collage like work where you're able to combine and stack it. And it's this, this modular system, which is

really at the root of what language is. I mean, authors like William S. Burroughs were playing with this concept back in the fifties when he would take a whole book and cut up each individual word and then slide them together in new combinations to make these new works. I mean, there's a lot of interesting things you can do when you look at language as a tool and less as this immutable thing that we were endowed with and must protect at all costs. It's like, well, we created language and playing with it in a, in a system that is outside of our brain is quite, it's quite interesting and refreshing, I think. Yeah. And, and it is how some would argue the best thinking happens is by actually playing with the language outside of your own brain.

As a serial note taker, I couldn't agree more. And I mean, we have, we have the kids do it already, right? When you ask them to come into conversation with the text, when you write in the margins of your book and you enter into that conversation with the author, you are playing with the texts, right? The ways in which we quote texts with ellipses and we're combining and shortening arguments that we already, we already play with it, but now we're kind of being able to play with this huge playground of texts that is quite frightening because it is, it is quite massive. I mean, it's, it's nerve wracking where the data came from, how the data was processed, what that says about us. It's, it's a reflection of our own language, which is, which is quite interesting. Yeah. Speaking of taking notes, by the way, if you read the book, how to take smart notes by Sonke Aarons. No, but I guess I will now.

Yeah. Put it on your list. It's going to be a, it's going to be a game changer. What is really cool about that process? Here's the, the gist of it. You take smart notes by anything you want to highlight, you write something that is related to that, where that is either the source or the counter argument or whatever. But the purpose is that when you take a note, you write what you think about it rather than just highlighting it. So in my doctoral coursework right now, I'm reading all these essays and research papers and all this kind of stuff. And, and my rule for myself is if I highlight it, that I have to comment on it and I can't just highlight it. So now I'm gathering all of these things, these snippets of text with my thoughts with it. And, and then at some point I'll be able to go through those. And as I'm

writing these other papers, be able to bring those things in and be able to go through the process much faster than going back and saying, what do I think about this highlighted text now? I already wrote down what I think about it. So now it's a matter of fitting that into the larger structure of whatever I'm working on. And, and, and that kind of strategy with an AI tool to say, here are my, here are the quotes and the sources. And here are my thoughts about each one and make this into a coherent flowing idea. It can be a quite powerful process that could save a lot of time. But again, you've got to put the work in upfront in writing down your thoughts from the beginning rather than just highlighting things.

I mean, artificial intelligence will be fantastic for managing large databases and the like. So as if a human takes copious good notes, you could do a lot of interesting things in making that, that knowledge base more active where you can kind of query it with something that can speak to you back in the, in the framing of your notes, which would be, which would be quite, that'd be quite fun to have now that I'm thinking about it. I mean, I have an app that I use obsidian for my notes. It's just all mark. Oh, okay.

There it is. It's just, I love being able to connect the knowledge base through those backlinks and using the system of tagging. I just, I find that to be a useful system and is different than how I would previously take notes, which is that kind of file format where you put things in different folders and the folders inside of folders and you number all of it. And it usually gets lost, but in making your knowledge base, something that is easily searchable and you can see the connections between thoughts seems to be more useful for me. And that's just a credit. I promise you, I'm not getting money from obsidian on this, but you know, I couldn't recommend it more, man. No kidding. So I've actually taken markdown notes since 2011. And so my obsidian library, and I've moved from several different things. I did Evernote for a while, but my big problem with it was that when I had to export everything, it was a major pain to get

everything out of there. So, so anyway, I've been taking notes like that for a long time. All of my podcast notes from every interview I've ever done are all in there in this format. And, and what's really cool is that there are some chat GPT plugins for obsidian. I saw that will, will allow you to start making connections that are, that are pretty cool. And this is, this is one of those ideas just to continue on this a little bit longer from how to take smart notes is that you have all these connected notes and you find ways to bridge them and connect them. And what I've, what I do when I read something is I almost always read it on the Kindle. And if I don't have it on the Kindle and I do have it in PDF, then I read it in the read wise app and then revise syncs to obsidian automatically. And when I take notes,

I will link to other notes because I know what they're called in the comments so that that link is already established, which is so nerdy. Yeah. So nerdy. That's awesome. But I knew you'd appreciate it, but yeah, but that that's what makes some of this stuff really powerful is that one, you, if you can start finding those connections automatically without you having, without you remembering that that was there, that's really powerful. And there've been a couple of times where I've been able to find some of these other notes. In fact, it just happened last night. Oh my gosh. I was, yeah, I was, I was working on something for my doctoral program and I was like, man, I swear I've done something around this before. And sure enough, I go find this note from like 2016, 2017, something like that.

And it was like exactly what I needed in that moment and totally filled in a gap that I had. And because I had this system where I could search through everything and find what I was looking for, I was able to connect it back fairly easily. Which is quite remarkable. I mean, first that's every note taker's dream, right? You, you come back and you're like, there's this note from this thing you took five years ago, six years ago. I just have a drawer of paper notes. So, because I've been so sloppy and noting until now, I only started using Obsidian maybe late last year. But that said, the use of AI in connecting those, those disparate notes in a way that you hope you would connect to be learned material with your, with your past prior knowledge, it's not quite the same thing, right? And understanding and exploring how artificial intelligence learns, how these large

language models learn, language models learn, and differentiating that from how humans learn, begin to reveal a little bit about the educational process. And I mean, it's common knowledge for all teachers. You know, you learn into relation to what you already know. And ways for using artificial intelligence to maybe in conversation with students gauge prior knowledge around a specific topic, so you can better address student shortcomings. I mean, prior knowledge has a huge, huge effect size on any kind of new learning. I mean, a lot of students approach a new topic without the requisite knowledge to do well in that topic. And then we're wondering why there's such variations in grades. And we try to do interventions at that topic level where it's like, no, no, you're missing the more fundamental, foundational piece. It becomes very apparent when you have a student who has a, say, much, much lower reading level than where they, than where most of their classmates are at a certain level, right? Say 11th grade and you have a

student who is measured reading at say 7th or 6th grade reading level. That could create huge barriers in them being able to come in contact with the material. And that, that's a very obvious example. But what about the more fine-tuned stuff? If you try to go into the American Revolution without the students having the proper nested context of understanding the British salutary neglect and the French and Indian Wars and the like. So then you have this, this requirement for prior knowledge. Being able to use artificial intelligence in a way that may make a student's lack of or their ability with various pieces of prior knowledge going into a lesson would have tremendous value. That's not something that EduAid does, by the way. I'm just speculating about where the technology could go. And that'd be quite interesting. We try to get there by making very explicit and very clear at the beginning of most of all of our generations that you need a certain piece of requisite knowledge and what that

knowledge would look like. So a teacher can go in with some knowledge there. But yeah, prior knowledge seems to be huge and the uses and implications of AI would be quite interesting in that domain. I mean, I can see a whole, you know, we talked about tutors for specific skills, but even having tutors to give the requisite background knowledge would be incredibly powerful. And I believe from what I've seen with kids who do have the background knowledge, that would be huge in helping kids be more successful in learning the things that they need to be learning. So this conversation has been awesome. I so appreciate your time and I feel like we could go on for another time flew. I thought it was 15 minutes in. I know. Right. So we did not even get to Ludwig Stein and Wittgenstein. Yes. And we may need to come back for that. I would like to come back for that after I've learned a little bit more about

Ludwig and, and get some, my own background knowledge. Yeah. That'd be a lot of fun. Yeah. So, so this was good. How, how would you like people to reach out and connect with you if they, if they want to, if they want to try EduAid, what's, what's the best way to do that? So EduAid.ai. Again, we're totally free. We're totally open. You make an account. All it takes is you type in your email address. We send you a link. You're there. No password necessary. Just get right in and we save a, we don't really save any data at all. Right. Nothing gets corded or anything like that. We should make that clear. I know the education space is, is, is very much FERPA and COPPA compliant.

Yeah. We are that. We never ask you to input any kind of personally identifiable student data or anything like that. So it's fine to just play around with and see what we have. But if you want to talk about EduAid, contact at EduAid.ai, or you could email me, thomas.thompson at EduAid.ai. Those are really the best ways. Instagram and TikTok, of course, and all the social medias were, were there. I personally don't really look at the social medias. That's, that's Thomas Hummel, the other co-founder. I'm very happy that he's savvy in that department because I, I'm not.

Yeah. That makes sense. Well, again, this has been great chatting with you and, and thanks for your time, Thomas. Thank you for, uh, for the, the thoughtful way you're approaching this. I think that's really valuable. Thank you. This has been great. Do you want to simplify your school's technology, save teachers time, improve students' performance on state assessments? You can do it all, but don't waste another minute. Head straight to IXL.com slash BE to learn how IXL's research proven teaching and learning platform can help you achieve all these goals. That's IXL.com slash BE. When I was a principal at a title one elementary school, we had Playworks and it dramatically changed how our school functioned. Kids show up to school when they want to be there. And that's the whole idea behind Playworks. What they do is they train staff to build belonging and connection right into recess. So the kids feel connected and want to come back.

They have tons of games that they teach the kids that are fun, enjoyable, and games where either everybody wins or the loss gets you back in the game very quickly. One principal put it, when the teacher hops in the game and everyone's playing together, that builds camaraderie between students and adults. And the research backs it up. There was a UC Berkeley study that found that how-to-one schools partnering with Playworks had significantly lower chronic absenteeism.

If you're thinking about attendance this year, and we all are, visit playworks.org slash podcast to get the 30-day attendance reset and additional helpful resources that you can implement today. Don't wait on this.