Make It Mindful: An Education Podcast

Seth Fleischauer (00:02.306)
Hello everyone and welcome to Make It Mindful, the podcast where we explore how to keep schools relevant by looking through the lens of mindfulness and asking the question, what's really worth paying attention to here? My name is Seth Fleishauer. My co -host Lauren Pinto is on a break and together we delve into the world of education by interviewing change makers and focusing on practical, transformative solutions for teaching. And my guest this week is Aaron Boffman. Aaron, welcome. Thank you so much for being here.

Aaron (00:29.452)
Thank you for having me.

Seth Fleischauer (00:31.234)
Could you start by introducing yourself?

Aaron (00:34.348)
My name is Aaron Boffman. I'm the assistant superintendent for instructional services in Northville Public Schools in Northville, Michigan. Former teacher, middle school principal, professional development director, and currently working in curriculum and instruction for the last probably 10 years.

Seth Fleischauer (00:51.714)
Excellent. And you and I were connected via Jordan Lewis, who's actually a guest on this podcast earlier. And he connected me to you because you're someone who's working, I think, at the forefront of schools adopting AI. You basically, you've got this district right now that is full of teachers who are experimenting with AI in the classroom. And you're even piloting several student -facing tools.

And I am curious, how did you pull this off? Because there are a lot of political headwinds, there are a lot of logistical challenges to doing what you've done. And I want to hear just how you've gotten to where you are at this point.

Aaron (01:28.524)
It has been an interesting ride. You know, when chat GPT was released last year, we started to hear a lot of outcry about blocking things and what we're going to do. And then in the fall, we just began thinking about how to get our heads around it. And so we brought in some trainers from Michigan Virtual who did some sessions for our teachers that just really laid the groundwork for it, including for our admin like myself, who were just.

trying to figure out what to do with this thing. We knew we didn't want to stick our heads in the sand. We knew we wanted to do something with it. And we saw it as a much bigger thing than anything we really had ever adopted before. So we began that training. And then I just became immersed in it myself and started using it. I'm married to a sixth grade teacher in a different district. And her and I worked together to try to play with some of these tools and begin to experiment. And really from November till January, I just worked hard to become as knowledgeable as possible.

And that's when we began conversations with the Board of Education. And I began doing some training with them and helping them understand generative AI, large language models, and the way that I think that this tool and these, really the artificial intelligence itself can impact our district. And our board is very forward thinking. I work for a superintendent who is very open to taking risks and doing innovative things. And so I just said to them, if you give me some space, I will figure out how to...

make this work for us and keep us on the cutting edge of things and more importantly impact the classrooms. And so we layered out some more training. We did a follow -up training with the board on student facing tools. Once those began to be released by places like Magic Student and School AI and CurePod. And then I began the process of actually getting teachers on board. And so we worked with them. We began training them on individual tools.

We pushed out some large language models that were a little safer, like Copilot and Gemini, where we were enterprise clients, so we were able to access those tools in a more private way. And then we began working with organizations around the state. So I joined the Wayne County AI Task Force, where we are located. I joined the state task force that's working with organizations around the state to look at AI. I began visiting and having conversations with all kinds of people. And more importantly, I started using the tools.

Aaron (03:43.596)
I was helping to develop lessons with my wife. She was trying them with her students. We were trying them with teachers in our district. And then we were getting real feedback. And so we were seeing what kids could do. And honestly, what we discovered was there are so many layers to it. But ultimately, this was a tool that can genuinely change the way we approach instruction and really for the first time ever feel like we can personalize and differentiate in ways that we've talked about for a long time, but we've never actualized.

And so this has really been a real push. And so lately, my last pieces have been really starting to share with parents. So notifications went out to them. We had built a board policy and procedure back when we shared with the board the training we did with them. And part of that was parent notification. So we have a parent letter that goes out to any classroom that's using this tool. And then I've been meeting with some PTA groups just to talk about it in a proactive way.

And really, I think that's what's been beneficial for us is that we've been really proactive about this. We've come at it from AI is good. AI has great opportunities for our kids. AI will be a requirement for them to understand. And yes, it has some things that can be negative. We're going to try to counteract those things. But far too often, we find the fear leading. And so people aren't able to actually get the work done. And I watched so many places committee this thing to death. And they just get further and further behind as a result.

Seth Fleischauer (05:03.714)
Hmm.

Seth Fleischauer (05:07.65)
Huh. Wow. There's a lot there that I want to unpack. And maybe a good question to start with is, you know, because I want to know about the impact. And you teased it out a little bit. You were talking about changing the way that we approach instruction, the personalization and the differentiation that's possible with these tools. You talked about engaging all stakeholders, making sure that you are not just talking to administrators and teachers, but also to parents.

In these groups, there are naysayers, right? There are people who are going to not want you to do what you're doing. And I'm wondering how you engage with those people. Is it about the impact? Is it about, you know, talking about that the fear leading and how a place of ignorance is not a good place from which to make decisions? Does it depend on the person? How are you convincing people who are not on board that this is something worth pursuing?

Aaron (06:07.372)
So my first naysayers were teachers, right? It was cheating, it was cheating, it was cheating. And so we showed them some ways, tools like Brisk that have some replay options so teachers can see like writing can be monitored better. We don't need to run to blue books and go back to in -class handwriting in order to protect the environment. But I also think we talked about changing instructional ways, like we gave them examples of ways you would change an assignment rather than just saying, write the essay, there are some other things you can do.

But we're still overcoming some of that ultimately because AI can really do most of the work that the traditional school system has been doing. It can do that for you. So we do have to rethink. But when I get to parents, we've been sending notification letters out and I've been following up with teachers. I haven't had a lot of folks who've said my kids not to be doing that. As a matter of fact, no one opt out. But I do have the questions that come up. And so what we try to do is give them a couple of things. One, we talk about the workforce.

We've been working with some individuals specifically, and I'm fortunate in my district to have people who are already seeing this. We talk about how it's already changing the workforce landscape. And if it's happening now, their students are gonna have to know how to use this. And so it's not a matter of most places that we've talked to about. If you'll use AI, it's how you'll use it. And so we tell them like, when your kids get out of college, even the seniors right now are gonna have to know what to do with this in five years.

They're going to have to do a lot of doing even in college. We've even talked to some folks in college and talking about ways it's already starting to change instructional models there. And then we also talk about the fact that if they want to be on the front end of this, they will give their kids a leg up. They will have a higher level of experience than the students in college with them because many of the districts are just blocking these things and putting their head in the sand door.

or just afraid of it and not knowing, or they have a really long process to go through. And because we are staying ahead of it, those students are gonna have a competitive edge. And parents, I'm in a high achieving school district, so our parents appreciate that, that their kids are going to have that competitive edge. And we share with them that we know they'll be dangerous pieces, but we have also worked really hard to make sure we're only working with companies with strong privacy policies.

Aaron (08:19.148)
And even with our student facing tools, there's no student facing tool where a student is actually logging in. That's how many of these companies have gotten around the FERPA and COPPA. And I shouldn't say gotten around, but they've managed those situations and they've been able to show us how they're compliant. I never put a tool in front of a student or a teacher that I have not vetted myself. I can't tell you that there aren't teachers using some of those tools on their own and trying some things out, but they're not certainly not being pushed out by my team. And I think that those conversations,

while they have been limited for me, and I'm very thankful for that, they've been pretty easy to overcome once they see the value and I show them what kids can do as a result of having this and how they will get a step ahead. That really helps to disarm them. And then I remind them always at the end, there are gonna be ways, even for them personally, that they're gonna have to be ready to handle it in their own lives, that there are some ways, just like we've done with everything since the inception of fire, we've found a ways as humankind to do everything.

to do something bad with everything we've ever created. And so they're going to have to confront those things. And I've tried to give them some strategies for that and conversations to have with their kids around using AI in the right times and the wrong times. But ultimately, that's still a lot in flux. And as kids get savvier and programs get better, you're going to have programs that will help them rephrase things to try to avoid AI detection. I mean, there's going to be a lot of things. So we really have to rethink.

Seth Fleischauer (09:18.306)
Hehehehe

Aaron (09:43.116)
instructional model more than just try to figure out how to keep these programs out of kids hands.

Seth Fleischauer (09:50.114)
Yeah, a lot there too. So with like you just talked about how these tools will likely evolve into a way where they might get prompted to rephrase it so that they get past the AI detection. Right now, there's so much that goes into prompting because these tools are not the best at guessing the context. But you can imagine that they'll get much better at that. And so do you?

have a concern at all? Like when you are, when you're arguing that these tools, like being comfortable with these tools are going to put students on the cutting edge, they're going to give them a leg up. Are you worried at all about the evolution of these tools and that we're perhaps teaching some skills right now that are going to be evolved out of the technology? Or do you see it? Do you look at it in a way where you think that the skills

that you're teaching right now around these tools will persist no matter what the evolution is.

Aaron (10:50.444)
I certainly think we're gonna find ourselves always feeling a little step behind because this stuff is advancing so quickly. I mean, I get an email every morning saying, here's the new tools for today. Like it's not like this week or this month, you know, it's seven new tools today. So we're always gonna feel a little bit of a step behind, but prompting to get to that really heart of it, I remind people that all of these tools are just different wrappers that are put around bots, right? And bots are only so smart right now, they get smarter every day. But how we wrap that bot, that,

Seth Fleischauer (11:01.602)
Yeah.

Aaron (11:19.66)
prompting or even these school tools that have been developed, or just people who've taken them and pre -programmed those bots and giving you some space to program your own bots, that's really where the value comes in. So if we can help kids see that these are robots that are being programmed, and the difference now is that anyone can program them. You don't have to have any fancy background to understand coding. I can program a bot to do just about anything, good or bad. And so we have...

putting tools in front of them that are safer, that we know won't do the bad things, that won't allow them to. That doesn't mean they can't access them on their own. I can't protect them from the rest of the world. But in our situation, we're going to teach them how to use these bots in a productive way. Teachers and students both need to learn prompting. That, I don't think will ever change. The difference is, as we go along, more prompts will be available for you to harness. There'll be prompt banks and other things which are already starting to develop where you're like, yeah, that's exactly what I want, but I'm going to change this word.

So prompting and what you want, what I've done in the last eight months with my prompting has been monumental. I mean, even this morning I was writing a prompt for something I need to do with an interview next week for some coaching jobs and I was realizing like, yeah, I need to say this, I need to say that because if I don't, and then I pushed the button, I was like, yeah, it's missing this and I changed the prompt and done. You know, like 20 minutes, I've got everything I would have needed that probably would have spent hours doing before. And it gave me everything I want for that interview. But I've learned.

how to really make sure that I'm detailed and asking what I want. So I think if we teach kids that, along with digital citizenship, which is shifting too, that's a whole nother landscape beyond what we've talked about digital citizenship in the past. I mean, I really feel like we as humans are going to have to begin to shift to everything is not real until proven otherwise. And I think that could be a little scary for people because when I look at things now, when I scroll through X or other social media platforms, I'm like, yeah, that's probably not real, right?

Seth Fleischauer (13:02.53)
Hehehehehe

Aaron (13:11.66)
There's a giant planet coming at us and it's probably not real, right? I have to assume that until I have other things to confirm that those are real images or videos and that that person's preponderance of evidence is that they might say something like that. You know, I look at the example with the Baltimore principal, who's athletic director, posted some things about him. I would hope that people would have stopped to say, does this sound like this principal that he would have done this?

Seth Fleischauer (13:30.754)
Yeah.

Aaron (13:36.556)
and that the preponderance of evidence would have been, something's not right about this, but because we're so early on, people don't know to be skeptical. And so I think we also have to teach them that. So the prompting, I think, will not change. The bots will get much more sophisticated, but they're always just waiting for us to tell them something. Now, when it starts to get more human, emotionally based, and understanding human thinking a little bit better, I think it will start doing some of that for us, and then we're gonna have to just either rein it back or tell it more.

But that prompting will never change because they want to know what do you want from me. The bots want to know what you want from me so that I can get you the best possible advice. They do want to make you happy. That's part of their planning and programming. So I think if we do that, kids will always get better and they will grow. We can't just say, though, like this tool or this thing. Like, I don't think much longer people will actually go to ChatGPT.

because it'll be built into everything they're using. Like there'll just be a large language model embedded into everything. Like right now my teachers don't need a large language model because school AI has a large language model embedded. As a matter of fact, it's really chatty PT with a different wrapper on it. They call it co -teacher. For teachers, they call it sidekick for kids, but it's a safe version that only has a one -way street and that protects their privacy. And we've middle schooled all these things up. Hey, tell me where...

Seth Fleischauer (14:45.826)
Yeah. Yeah.

Aaron (15:00.364)
how to bury a dead body, tell me how to get away with using a lot of inappropriate terms to make sure that it won't respond to those things and it doesn't. And that's what we're really trying to make sure of as well, is protecting, putting some gates up at our schools, like we always have, to try to make sure our kids have a safe place to go so that they don't feel the need to go to chat GPT and find themselves in places they really shouldn't be as 12 year olds, where they've got a safe place like school AI or magic student or other things to really do that in. Even our.

Seth Fleischauer (15:03.042)
Hehehehe.

Aaron (15:28.524)
our kids in Co -Pilot can have a safer environment because we're enterprise clients. So if they're logged in with their school account, you know, they have those things.

Seth Fleischauer (15:37.186)
Yeah, you bring up a lot of interesting points here. The digital citizenship piece is enormous. I basically just see a lot of those practices being supercharged right now. Like you said, you know, everything is not real until proven otherwise. I think that that was true of text for a long time and now it's going to be true of video and it's going to be true of pictures, right? These deep fakes like the one with the Baltimore principal who, you know, his AD put together, used his real voice and put together a

a clip of him saying some horrendous things and then put that out there and I think he's still he's like in limbo right now which is crazy. But so that that's a huge issue. You've teased out a couple of times the instructional design piece and how this is impacting that. You talked about how initially when you were trying to convince teachers.

You showed that brisk is something that you used for writing and how that impacted. You gave examples of how that impacted the instructional design and how you're approaching it differently. I'd love to dive into that a little bit. I know that you've done some of this work with your wife and experimenting in her classroom. If I were to use something like brisk for writing, how am I making sure that students are not bypassing the cognition that they need in order to become better writers in order to become better?

evaluators of what they're getting out of the AI. How do these tools and how do these different approaches work in order to make sure that students are still thinking? Can you give us some examples?

Aaron (17:06.476)
Yeah, yeah, I would say that the thing about AI is that there's a hundred entry points for it with teachers. There's everything from like simple things like joke generation and, you know, and leveling text, which is a really easy entry point for a lot of folks to try it out, to then taking that and building actual lessons and content and all the way to changing the actual way that instructions happen in the classroom. And I'll share a couple of examples. So I start with things like brisk.

What Brisk has done is worked with Google to basically take their editing history, their revision history that we've been able to access forever, but it's turned it into a video and added in some icons. So as a teacher and as a student, because we're going to see Brisk Student, which is called Brisk Boost, come out in a month. What you see there is the teacher can push play and watch the students writing process from beginning to end. They can actually watch every word, every edit, every correction in an actual loop.

instead of having to click on each revision in Google's history. And what that does for the teacher is quickly be able to tell when things large chunks were pasted in or that are not cited, you'll be able to say, where did this come from? Or if it's all yellow, and it also gives them timing, like how long they spent writing it, when it actually has a timestamp on it, their entire revision history. So you can even tell like how every single second and minute they worked on it. And then you are able to in brisk give feedback pretty quickly using AI.

Seth Fleischauer (18:03.682)
Yeah.

Aaron (18:29.676)
And then you can watch that revision history again to see how they responded to. So not is it just an AI kind of detection tool that's really authentic, because I'm not trying to depend on a computer to figure it out. I'm actually watching it. And it has little icons where you can click on it. It will show you anywhere there was a paste. And you can see those pastes and see if there's simple lines that are going to be cited later in maybe the work cited page or footnote. But that has really been helpful. Teachers love it.

It has a lot of other things too. Those are the two main components that I think set it apart from other tools. So that's a really good way for me to say to teachers, hey, why don't you just try Brisk? Like, it's not really AI in the sense that it's not changing instruction, but it is giving them some idea about, hey, look, I can actually watch kids' writing process and see their thinking. If I've ever collabed with a person in a document, sometimes I just like to watch them think, like actually typing, just to kind of.

get into their head. Well, this does that with every single kid with every single essay and it does it instantaneously. I don't have to do anything except push inspect and it just goes through and does that for me. So then we move into like tools that are pretty commonplace now, which I know sounds crazy because eight months ago we would have been so impressed by this, but where they can level things for special ed students and students who are struggling and students at the advanced level.

So before we would have to give different text to different kids because it was written different levels. Well, now every one of these tools has the ability. So I have some teachers who are in our special education department who are using tools like Magic School and Magic Student to provide the same text to kids at the level that's accessible for them or even sometimes just slightly above. But it's meaning that now that text is completely accessible to every student. And so it's the same concepts, same topics. And with the tools we have, it doesn't even have to be in digital. Like you can take pictures.

of a page and have it rewrite that page. You can upload a PDF. I can write it a Post -it note and do, I mean, it will take anything and it will level it to whatever lexile you want. And so they're finding their kids actually being able to access subject -based material like science and social studies, but at a level they can understand instead of having to try to figure out alternative texts for them. The other thing is I've got some folks who are writing using...

Seth Fleischauer (20:38.05)
Hmm.

Aaron (20:41.708)
tools in some of these that generate informational text. I'm gonna elementary teach you one of the hardest things I remember doing was trying to find a narrative text that would pair with an informational text. It's easy to find narrative, it's harder to find informational, and now these generators, you can simply say, this is the story I'm doing, can you generate an informational text around the same topic? And it will generate text, and it's not gonna win a Pulitzer Prize, but it is enough for me to be able to do that part of like compare and contrast between informational narrative around the same topic.

Seth Fleischauer (20:51.714)
Hmm.

Aaron (21:10.636)
or simply some of the tools that allow you to take a YouTube video and put it into a link and it'll do lots of things with it. It'll create stopping points with questions and make it interactive. It'll turn it into a text summary. It'll give multiple choice. It'll pull the vocabulary. So those are kind of mid -level things where you can change instruction that way. And then there's the really advanced stuff, which I think eventually will become not so advanced, but right now it is certainly, and it's not for the early adopter. It's not for the person who's just figuring this stuff out, but.

One of the programs we built with School AI with my wife was an ability to run an entire lesson. We did it twice, once with CurePod, once with School AI. Both times very successful. My wife is in a district with a lot of Atris students and students who are oftentimes behind academically. And so we built this program and what it did was it gave kids the ability to communicate with a bot the entire time. So she launched the bot, they joined in, they were on their Chromebooks, they were communicating with this bot and we had programmed it.

to basically do a writing lesson with them. And we gave it what our expectations were. We uploaded a link to a document that had the outline of what we wanted in the writing. They were working on a book. And so we were going to connect it to the book as well. So we did all of that in the prompting. It probably took me 25 minutes to write the prompt to really get it where I wanted. It was kind of a prompt and then preview prompt and then preview. But what she did the following day, which was amazing, in second hour I got a text from her saying she wanted to kiss whoever invented School AI, was six class periods.

Seth Fleischauer (22:35.714)
I'm sorry.

Aaron (22:38.06)
Every single kid was able to get through an entire rough draft with feedback by just following the prompts and hearing from the bot. And what she was doing is watching their chats on the screen and seeing where they were at, because it will tell them so -and -so is struggling, so -and -so is doing well, this is where this person's at. She could see their chat logs. And she's walking around the room doing writing conferencing with individual students. But everybody else is engaged.

It's the thing we've always wanted for that kind of reading and writing instruction, but now it's real because she has 30 assistants in the room with her. And so by the end of that period, those kids followed through, built their first, built their second, built their third, and built their fourth. And we told that things like, you know, a lot move on until they prove this and that they're able to do that. And we did that with School AI on the one day. We did it with Curapod, which is more moderated where teachers kind of leading them through the process. Same kind of thing. They worked on a chapter summary.

But she was able to push a button. They got feedback. They rewrote it. It gave feedback on their new rewriting. They moved on. Stuff that she couldn't do without a great deal of time. And all of her students left that day with a rough draft. And now they take that rough draft, they put it in Google Docs. She can put it through something like Brisk and give feedback. You can see how that just changes the way that she's able to do instruction because now she can actually confer with kids who are struggling. The kids who are not struggling are getting what they need.

and the kids who were advanced are moving ahead. I won sixth grader, I was playing with this bot forever. I wanted it to just take all the pieces and put them together so they could just paste it into a Google Doc. I was late and I was tired, so I just said, you know what, they can just copy and paste. The sixth grader in one of her advanced classes was like, hey, can I ask the bot to just do this for me? She's like, hey, knock yourself up. But if you do, don't tell my husband because he couldn't figure it out. So she does, she figures out what the question to ask, the bot basically takes all the pieces and moves, and then she just pasted in one rough draft.

Seth Fleischauer (24:24.737)
haha

Aaron (24:32.396)
You know, we were able to do two different bots for advanced versus her general ed classroom Like we asked we asked a little bit more of the advanced students to push their thinking so we asked them to do additional things in their writing So we just modified the prompt and built it as two separate so you can see there's a continuum here of like yeah email responses and and just you know joke generators and song generators like that's what I kind of get people's interest doing and then you move into what if we were to like

bring in actual materials and do something different with it. Even if we layer it in, like I was in the teacher's classroom, we were doing the Boston Tea Party. She was doing the Boston Tea Party. And I said, you know, we could build a bot of a person who's on the boat from the British side whose tea is being thrown in the harbor and someone who was at the harbor throwing it in and they could ask questions and they were actually having these real interactions. Like they could just layer that into your lesson, right? You don't, it doesn't have to change the whole lesson, but now you.

put kids into two teams and this team is team Britain and this team is team the American Colonials. And you now have them having a conversation about the difference because they're talking firsthand because it feels really firsthand like you're talking to an actual person on the ship or an actual person throwing the tea in the harbor. And while it's not perfect system, it will get so much better. And it's enough of the information. I don't think I mean, we have a whole conversation about bias and other things we have to talk about, but it also gives them a perspective like, you know,

Maybe the British don't see the Fourth of July the way that we do. And so it was a really cool way to just add a, just to snip it into. So you can also layer those pieces in. And then of course, there's the ultimate dream of how does it change the entire instructional model? And I'm not talking about just putting kids in headphones in front of screens, because there's group work that can be done. There is teams that can be built. I think that's essential, but it's changing the way the teacher can now interact with students and getting so much more done in such a quicker fashion.

Seth Fleischauer (25:58.018)
Yeah.

Seth Fleischauer (26:27.618)
Wow, this is why I want to have you on the show. These examples are just incredible. I do want to add just a point of clarification here. You're talking about programming the bot. You're just talking about writing a prompt, right? Like that is the programming. And I've seen this before, these prompts that are essentially like, you know, do this first, wait for this thing to happen, then do this. Once they've done that, once they've shown something like this, then you can move on to this. And so you're essentially like, you know, planning out.

Aaron (26:30.636)
Hahaha!

Aaron (26:40.428)
Right. Right.

Seth Fleischauer (26:56.258)
the learning journey of the student in a way that you might when you're conceiving of a lesson plan, but you're writing it in a way that turns any of these chat bots essentially into a personalized learning software, which blows my mind, right? That you can just, that it's just a matter of using the right words that instead of computer code now, it's just describing what you want it to do. And...

That potential, I think, is so enormous. You said it's advanced for now, and I agree that it is advanced. But I think that, more than anything, has the opportunity to really transform education, because we've seen these personalized learning softwares do that for education. We've got an entire generation of students who've had content put in front of them that's actually at their level.

for the past like 10 years or so, but now you've got this extra layer of like, no, this is exactly what I'm working on in my class right now. So I don't have to be dependent on what has been developed at market for personalized learning software. I can create my own with the know -how of programming these in the right way. And I just think that's incredible and it's amazing that you're on the forefront of this. You did mention,

bias and digital citizenship and you know you're like it's not perfect right like those though that truth is out there. These models are trained on the language of the internet. Last time I checked the language of the internet did have some bias in it just just a smudge right. So so I'm wondering like with all of this excitement with all of this potential you've mentioned it a few times that has to be part of the conversation but how do you bring it in.

How do you teach digital citizenship in the age of AI?

Aaron (28:53.036)
Well, there are some places that are helping try to do some things like Common Sense Media has some pieces out there in digital citizenship. I was presenting in Varian Springs recently and one of the education consultants from Common Sense Media was there and she shared some of that curriculum. I think what we have to do is find ways to embed it into the just the normal way of life that we do everything. It can't be just in a tech class or just in this special class we teach. It has to be part of our conversations always. One of the things that...

we also can do is begin to intentionally give them biased information and ask them to begin to critically think about it. One of the ways you can do that is potentially ask about, to write something more biased, less biased, and non -biased. And that's gonna require some prompting in your part as a teacher to say, yeah, this still has a little too much bias and it changed this, changed this. Of course, we also have bias in our own, right? We carry our own bias even from what we're reading. So I think you can force that in front of kids and say,

using a critical eye and a team of people around you, let's talk about what you see in here that needs to be reflected. Now, in that room, you have bias, right? Even within your students. So even them looking at it is gonna come across with some conversations that can sometimes be difficult to have. But I think we have to just continually ask the question of what do you see? So if I put in, ask them to put into a bot three different, and go to Co -Pilot and.

Gemini and Chachapi tea or school AI spot and just put in the same prompt to see them all come out and then even think about like what's different about this one versus that one? What do you see that's that's not aligned or even ask them to go out and find in the old -fashioned way some resources that will defend or refute points in the Generated stuff so that it trains them on that model of like how to find and how to look for Thinking of again about preponderance of evidence. I had a parent

one of those really early conversations, she said, like, my kids are on here, are they going to learn that the earth is flat? Is it going to be in there? And I said, well, you got to remember it's pulling from millions and millions of things, right? And so it's going to look at preponderance of evidence. So I actually did the prompt for her and then I screenshot it and send it to her, what, chat GPT early on, 3 .5, right? What would it do? You know, I understand the earth is flat. Can you tell me why that's true? And it came out and said basically, like, there are some folks who feel that way, but the evidence is such. And then it's shared in all the things.

Aaron (31:13.26)
And so I just said to her, like, it's probably not going to go that route because that's not where the preponderance of evidence is at, but it will identify that that does exist as a theory, an idea about some people. It's usually pretty kind about it. It's not going to say, you know, this is stupid and this is dumb and don't listen to this. It's going to just say that, yeah, that does exist, but here's what we believe to be true based on the evidence that has been permitted. So I think there's lots of ways to approach it, but ultimately we got to just put in kids' faces and say, there's bias in the world, right? Like,

You're going to, even if you just watch right now, watch two news channels, you're going to see the different reporting on the same subject. That's going to happen when you ask it things, especially academically, because there are opposing viewpoints. I mean, even around AI, if I ask chat bots about AI, it can often present the area of people right now who say this is dangerous, we shouldn't be doing this. And then those people who believe it's the best thing that's ever happened to the world. I mean, even if you ask,

even some of the forefront leaders around this and what they believe. Some of them believe it's going to lead to the extinction of mankind and others think it's going to be the greatest thing that's ever happened and we're going to cure cancer and all kinds of other things as a result. So we just have to force kids to have those conversations. I think that's really where that critical thinking and group work comes in because you can't be in an echo chamber as a student either because you are.

coming at it with your own bias, even as adults, when we have these conversations about AI, like I'm obviously on the end of people saying we need to be embracing this while I've got people on the other end. And I have to hear those concerns and think about, okay, well, maybe I'm coming at this because I've had so much exposure to it and I've done so much with it and I have to understand that they haven't. And so there's some fear and unknowns and misconceptions that I have to help bridge that gap. So I think we do all of those things because that, I mean, as you said,

There's really no form at all right now that doesn't have some sort of slight or bias one way or the other, even in places where it feels like it should be totally factual and third party, we know that it's impossible because humans are the ones who wrote it.

Seth Fleischauer (33:18.914)
Yeah, I mean, bias exists in everything, right? And what you're talking about really, I had a guest on the podcast once who was talking about media literacy and media, not just in the digital form, right? In all forms, like everything that gets put in front of you is media and there is bias in all of it, right? And because the bias is kind of supercharged in this situation, because...

So many of these issues are supercharged. I feel like it is this opportunity to like take a step back and look at it with a new degree of objectivity. You know, we talk about mindfulness on this show and that process of just trying to take just that little bit of distance between what's happening and what you're feeling in the moment so that you can be a little bit more objective about it. And I find that the technology.

is providing us with that opportunity. And in that way, I'm excited and optimistic. I too am a tech optimist, and so I see the potential. And I think that if and when we handle it well, that it can lead to some really great things. But there's that whole question of handling it well. You mentioned earlier the right and the wrong way to use it. I'd like to ask a question about policy and how you're dealing with that in your district.

What is your framework for how to establish it? Is it that you're letting the experimentation like inform the policy or is there room for experimentation? Are there very strict guidelines? Like how do you think about policy and approaching it?

Aaron (34:51.66)
So we worked with our legal team on a policy, which obviously is very, very early on. And we're kind of taking a little bit of a risk. But we built a policy that, so in our district, policies are generally very brief, and procedures kind of spill out more. So in our policy, it just simply says that the district will access artificial intelligence, generative AI, large language models under a carefully monitored exploration.

And we explained to them that this policy and procedure will change more than anyone has done in the past and more frequently than in kind of more narrow gaps. Like even for the fall, I imagine we'll have to modify it again based on where we go next, which is like a set of individually notifying parents, more of a district notification. And then the procedure was built, which really outlines ours in particular says the assistant superintendent of instructional services, me, will be responsible for kind of every.

previewing and vetting these tools, providing training, notification to parents, and really being the one monitoring these tools, but understanding that they will be used as an exploration that also likely to be needed to be changed for the fall. So it's not an exploration, it's more of a use case kind of situation. And so right now the policies and procedures are pretty clear and AUP, while they're not stringent, they do lay a lot of the...

pressure on my shoulders to make sure these things are right, which is appropriate. So no tool gets sent out from me without a full vetting of really time and energy, playing with it, using it, building in it, putting it in front of a few kids whose parents have been notified and making sure that it does what we think it does. And then ultimately mass distribution to anyone who wants to use it. So right now we're doing with nine tools.

And we will likely revisit that policy and procedure for the fall as we think about notifying parents in broad that AI is here and it will be used in your classrooms. And we will do everything we can to make sure privacy is protected and that your students are safe and that their information is protected. And so we don't use any tools that we don't have clear FERPA and COPA compliance on that are any student facing tools have to be one way street that that information has been kept. And we've been doing this for years. I mean,

Aaron (37:13.356)
Come to NWA and Edpuzzle and others, they've had access to our students' information too. They just aren't sending it out. So we're only using AI tools that do the same thing, only the ones that won't send them out. And they've done a great job of this by creating a microcosm that teachers build and students access. Students aren't logging into SchoolAI. So our policies and procedures indicate that we'll make sure that students' privacy is protected. And I'm doing that by only using tools where kids are not creating accounts or logging in and sharing personal information. And...

If they are having conversations with a bot, that information is being housed at that company. It's not going back out to OpenAI. And that costs them a lot to not share information. They have to pay a price to do that, because OpenAI wants all that information. And so we even found recently that with the opening of 4 Omni for users, when that occurs, or as that continues to roll out for free, a lot of these companies are actually seeing the prices coming down because of what they're paying to OpenAI is going down.

And so we're starting to see even some of those price shifts occur. But that's because they've been paying so much to protect that privacy and to create a filter that comes in so that it won't tell them how to bury a dead body and it won't use inappropriate language or accept conversations about inappropriate things. It will kindly say, you know, I think we should probably talk about something else or I think we should go in a different direction. And so our policies and procedures are fluid. They're relatively generic right now.

They will get more specific, I think, as we start to realize, like, hey, maybe we don't need our kids accessing ChatGPT anymore, and we can block it because they have other avenues. And why would we put them out there in the Wild West when we have a safe environment for them to plan? Most of my kids have data plans, so they can access ChatGPT if they want. I can't control those elements anymore than I can control their access to Instagram or Snapchat. But I can provide them a safe place and remind them that this is a better place for them to do this work where they're in a more protected environment.

Seth Fleischauer (39:12.866)
Yeah.

That's makes a ton of sense. You're talking about a one way street of data, copa, compliance. You're testing it for its efficacy, making sure that it says what it does, what it says it does. You're not companies that are not sharing information with open AI, making sure that it's not talking about inappropriate things like that definitely defines that walled garden. You were also talking about like no accounts, no personally identified information.

But earlier you were talking about an example where a teacher was monitoring like the writing of students as they go. That sort of implies that there was an account or were they just doing it by walking around the room and looking at their devices? Or is there a way to sort of have your cake and eat in two of like, you know, they have accounts. I as a teacher, I'm interfacing with their accounts in the way that I do a lot of other EdTech tools, but it's still safe and adhering to all your policies.

Aaron (40:08.556)
Yeah, so all of them use kind of a Kahoot model or a Pear Deck model where the students logging in with a generic passcode or a keyword of some sort. So when they log in, it then says, enter your name, right? And so they just put in Michelle or Jen or Frank. And the only reason you'd have to do anything different is if you have two of the same name in your classroom. And those are the names that show up for the teacher. But in everywhere else, they're giving a different anonymous name, like little tree or, you know, it gives them an anonymous name.

Seth Fleischauer (40:31.106)
Got it.

Aaron (40:37.42)
So the only person who sees their name, and I recommend for teachers, is their first name, unless you need a second initial for some reason. And again, that information is still not leaving School AI or Magic Student, like it's held within there, but the teacher has a running record. The kid cannot access that record again. It doesn't exist for them. So their chat log is gone, which is why I had them copy and paste it into Google Doc when they were done, because they can't go back in and see their work. The teacher can.

And she can reopen the session and let them re -access it if she wants, but they are locked out of it when she ends the session and they can't log into their Magic School account. It doesn't exist. So they are there for her. So, you know, kids can put in weird names, but she's going to say, okay, well, whoever put themselves in a Roger Rabbit needs to change their name, right? Like, I need to know who you are. And she, and they're, and you know, as long as they know she's going to see everything they're doing, because that's what I would recommend you doing is telling them.

Seth Fleischauer (41:14.946)
Got it.

Aaron (41:31.98)
I can see it all, then they're going to put their name in eventually, even if they try to be smart the first time. If she looks through her list and says, Devin, I don't see your name on here, what did you put in? Then she can do that. But she's the only one who's seeing that. And that information, again, is being housed. She always has those chat logs to go back to for months and months and months. And so I think that, again, it's housed there, but no different than a Pear Deck would have been.

when we used it, like the teacher has access to all that information, the students are getting a copy of what they did if you ask it to do so, but it's being housed at Pear Deck and they're not sharing that information with third parties, or at least that's what they said they would do in the agreement that we signed. And I would think that there would be a lot of financial obligations if they broke that privacy agreement. I don't think any of them want to be confronted with that.

Seth Fleischauer (41:59.138)
Yeah.

Seth Fleischauer (42:21.086)
Yeah. Is Roger Rabbit coming back with the kids? Is that popular?

Aaron (42:24.684)
I don't know where I came with that from. It's an old term, but usually they say something even more clever than that, but they use some term that nobody in over 40 knows.

Seth Fleischauer (42:34.402)
Yeah, yeah. So last question, and it's a theme that we explore here. People come on this show with some deep experience in a certain sector of education. And typically that experience provides them with the opportunity to see things a little bit more clearly and to notice some things that are happening that other people might not notice. And given your experience working with these tools,

Given the experience you have working with teachers and with students talking to parents, I'm wondering if you can share with us like a surprising observation, something that people might not think of immediately when they think of this overall landscape and dynamic.

Aaron (43:22.508)
That's a tough one. I mean, I've had a lot of surprises in this journey. I think what has been more surprising to me is there are some of the teachers who have really been struggling for a long time at how to do some of these things and watching them discover it and use it for the first time, because I do a lot of even small group sessions where we kind of play with a tool and I ask them, what's the outcome you want? And then we talk about what

tool might work. I'm now having people who are coming to me saying, what tool can I use to do this? And so people, I think, for a long time who have been struggling in silence and not wanting to ask for help, not only do they have a bot they can now ask for help. I can have a conversation with it about a curriculum issue. Like, hey, I need an activity on. Or this is this bombed. What could I do to make it better?

They actually have a one -on -one kind of coach right at their disposal that's completely non -threatening. But I also think for them, they're starting to see like, man, the things I have been struggling trying to do, small groups and other things, the one -on -one conferencing, are things I can now do. Like my wife, you know, she texted me and was like, I want to kiss the person who made this program because she can't believe how well classroom management has shifted and how well learning has occurred and engagement with kids who are not easy to engage sometimes.

It has done that for them. And so I think now that I'm starting to see people say, hey, I watched so -and -so do this thing. Could you teach me how to use that? That's been really powerful. And it's in some cases one heart and mind at a time. I'm not saying that my entire district is converted. I've got 40 or 50 teachers who are seeing the value. And that 40 and 50 turns into 70. And eventually, it'll really start to take hold.

But I think for me, it's just being able to see. I had one the other day, and it's just an example, a teacher who I'd never really hear from often, and I'd been in the district for seven years. She reached out to me and she said, you know that program you showed us where we could build a song? I had a new student coming in from France who didn't speak much English, and our students and I built a song for her, an AI. They use Suno.

Aaron (45:38.22)
and basically wrote it for her so that when she came in the first day, that's how they started the class with her. And she said, I was just in tears to see that this kind of thing could happen. And then as a small entry level piece. But for that teacher to see the power of what she can do instantaneously, I'm talking 30 seconds, a song is built and her kids get to see how they helped create that and make it come to life. And this is a program where it doesn't just give you words, it gives you the music and the singing like it's an actual song.

And when you watch that kind of happen and then you think to yourself, if this is possible, what else is possible? That's what the moment when the creativity starts to spark. And then imagine when we do that with kids, right? I've got a teacher in a marketing class who's using Canva Magic Studio with them. It's changing the way she teaches marketing. Her kids are bringing to life creations that would have taken them so much longer and so much more skilled to do in the past. And as marketing students, they're seeing like, wow, I can think it and create it in moments. Like,

Seth Fleischauer (46:17.538)
Hmm.

Aaron (46:37.388)
I can take whatever is in my head and turn it into reality by just asking the right questions. And that's, I think, where it's the most powerful thing for me is that moment of like, if A is possible, imagine what we could do with B and C and D. And they just keep pushing and pushing and pushing until we're doing things with it that we never could have conceptualized. People will call me and say, hey, you know, I found this tool. It'll do this. And I'm like, yeah, it will. What do you think you could do next with it? What do you mean next? I'm like, just think bigger.

Just keep thinking bigger because it'll just keep getting bigger. And they're like, well, what do you think? And I just ask a few questions. They're like, yeah, I'm going to try that. And yeah, do it. Let me know. Hey, I want to know how it goes. Let me know because I would love to see the outcome. I've got meetings scheduled between now and the end of the year to sit down with some teachers who want to learn. And I'm not telling them learn all nine tools at once. That's a horrible idea. I'm asking them, what do you want to do? What do you want to do differently? What do you hate doing that you'd love to get rid of?

Seth Fleischauer (47:27.682)
I'm sorry.

Aaron (47:32.94)
And let's see if we can find a tool that'll do that for you. And when you do that, they're like, man, if it can do this, could it do that? Probably. I mean, all we gotta do is try. I was sitting in a workshop one afternoon. I was doing an AI panel with some teachers and one of them was talking about bad practices. Like, we've sent 20 problems home with kids and we'd ask them to do them at home. And then if they didn't know how to do them, they just did 20 problems wrong. And now I gotta undo all of that the next day. And as she's talking, I'm in my computer keyboard typing.

Seth Fleischauer (47:33.122)
Yeah.

Seth Fleischauer (47:40.066)
Yeah.

Aaron (48:00.108)
And I said, here's what I want to tell you. I can even take a bad practice and turn it into a better practice, right? I just programmed a bot while you all were sitting here. You just watched me. That basically took those same 20 problems and said, do one problem at a time. Don't let them move on until they get this one. Have them try at least twice before you give them help. And if they do longer than X, then you need to give them the answer and teach them how they did it. All of a sudden, I just turned that really horrible practice into something that at least is a better practice, right? Somebody's at home as a tutor for them.

They're not doing the problems wrong because it won't let them leave until they move on until they've done it correctly or at least tried as many times as possible. And as a teacher, guess what? I now have a record of how they worked on those problems, how many times it took them and insights because most of these programs use AI to say like Terrence doesn't get it. You know, Julie understands this completely. I'm going to know those kinds of things instantaneously. Like before they come to class the next day, I'm sitting at home at night and I'm watching this happen. So even bad practices can be made.

significantly better by using some of these tools. I think those are the things that I think we start to step into. I wouldn't start with what I did with my wife with anyone on the first day. It's not a great place to start. Like that's overwhelming. But because we started with her using Brisk and doing some simple things and then moving into these other tools, that's how we got to where we're at. And now, you know, if she has access to those tools, what she can do with them is pretty exponential.

Seth Fleischauer (49:27.426)
Yeah. And you know, you mentioned fire earlier, like humans have been doing this with tools since we started using tools, right? It's like, what can we do? But what can we do with the tool? wow. We can do more with the tool. And this is just the most multifaceted tool I think we've ever had. And that creative piece of like, not just like, what can I do with this tool that is within these parameters of the established purpose of this tool?

But what can I do with this tool that is way beyond that that nobody else has done before because we are all in a phase of active experimentation with this. I mean, what an opportunity for creativity and what an opportunity to do things that you were previously limited by your technical ability to do them. You're no longer anymore. Right. Like it's it's it's incredible. Well, I think the work that you are doing, Aaron, is so fantastic. I would love to keep in touch and have you.

share these lessons that you're learning from this like active experiment that you're running over there in Michigan. I just think that you're, you're thinking about this the right way. You know, you talked about fear, like the antidote to fear is information and you are providing that for people. And I just, I have huge respect for, for what you've been able to pull off so far and I'm excited for what you'll be able to do.

Aaron (50:41.356)
Thank you, I appreciate it. It has been a fun ride. It feels crazy that this has been eight months, but it really is fantastic to watch what it's doing for teachers and for kids. And just excited to see what's possible in the future and just continuing to try to stay one step behind of ahead, because I don't think we're ever gonna be ahead, but doing our best.

Seth Fleischauer (50:58.946)
Yeah, yeah. Well, thank you so much again for being here. Last thing, is there anywhere that you'd like to send our listeners on the internet to find your work?

Aaron (51:13.324)
I do not have any place, I just work. So I don't have a book or a blog or anything else. I just do the work every day. So this is just a practitioner who's doing this thing and trying to help kids grow and be better.

Seth Fleischauer (51:16.034)
Hahahaha

Seth Fleischauer (51:28.258)
Yeah, well maybe that's why the information was so rich. Well thanks so much again for being here. Thank you as always to our editor Lucas Salazar. For our supporters, if you would like to tell a friend, follow us, leave a rating or a review. All of that helps the algorithm, especially the friend telling. That's the social algorithm. And remember that if you want to bring positive change to education, you must first make it mindful. See you next time.

Aaron (51:31.212)
You