Make It Mindful: An Education Podcast

Seth Fleischauer (00:00.654)
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 cohost Lauren Pinto is on a break and together we delve into the world of education by interviewing change makers and focusing on practical transformative solution, practical transformative solutions for teaching. And this week our guest is Lee Ann Scherer.

Share share did I get that right? Okay, sorry twice already in the first minute

Leigh Anne Scherer (00:32.462)
Yeah, you got it. Let's see yours.

Seth Fleischauer (00:40.654)
And this week, our guest is Leanne Scherer. Leanne, welcome and thank you so much for being here.

Leigh Anne Scherer (00:46.254)
Thank you. Happy to be here.

Seth Fleischauer (00:48.238)
Could you start by introducing yourself to our listeners?

Leigh Anne Scherer (00:51.63)
Sure, I am currently the technology director for a district just outside of Portland, Oregon. We have about 17 ,000 students and I've been doing that for about three years now. So, yeah.

Seth Fleischauer (01:06.478)
Excellent, yeah, those three years. Nothing happened in education the last three years, right? Just smooth sailing.

Leigh Anne Scherer (01:09.198)
No, not much happening in technology. Yeah, it's just been pretty quiet, pretty regular. Yeah.

Seth Fleischauer (01:13.934)
Yeah, yeah, normal stuff. Yeah, cool, cool, cool. Nothing to talk about there for sure. So we met recently at a conference put on at the University of Portland and you were presenting on AI policy, which is something that I'd love to get into with you here today. But I did also enjoy your survival story of the pandemic, especially given the fact that you stepped into your role. I think it was like, like,

Leigh Anne Scherer (01:19.63)
Thanks.

Seth Fleischauer (01:42.51)
right after the spring of 20, right? Having made it through that, having survived, looking back at what you did and what you didn't do, what would you say are like the lessons learned from someone running technology from the district level at a district in Oregon where you were, you know, helping these kids?

Leigh Anne Scherer (01:45.39)
Yes.

Seth Fleischauer (02:05.358)
helping these teachers and classrooms transition in and out of and around this new virtual learning and back into the classroom. What did you learn?

Leigh Anne Scherer (02:15.63)
Gosh, yeah, I started in technology in July of 2020. So that was a little bit of a wild time. At that time, we thought, when I started, we thought we were gonna be going back to school in the fall, that we would be back with kids, that we would, it would look different, but we would be back. And mid August, I think it was, you know, everything runs together now. Mid August, we heard no, we were gonna all be working from home, that students would be home, teachers would be home.

and we wouldn't have any in -person learning happening. So we quickly had to sort of develop a teaching model, develop schedules, make sure that all of our students had the right technology. We weren't one -to -one as a district at that time. We had almost enough devices for every student to have one, but it was a big mishmash of different types of computers and different.

You know, some were really old, some were really new, some were touch screens, some weren't. So we started with all of our students had an option to choose one of our devices, basically. And so we had to get things out to students. But then we had to figure out how do you work with students when they're on a lot of different types of devices? How do you help them troubleshoot over online?

how do you teach teachers, some of whom had really never even been on a Google Meet before, how to transition all of their teaching online. So it was very challenging. I think one of the things that we kind of joked about was our next pandemic, we're gonna know we gotta do this. Our next time we all get sent home, we're gonna know how to do this.

Seth Fleischauer (03:58.414)
Hehehehe

Leigh Anne Scherer (04:03.918)
But it was really just sort of taking it day by day and figuring out what was going to work, what wasn't going to work. And at the same time, the technology would change. I don't know if we talked about that before, but that was another thing that we were grappling with is we're a Google district. So we've been in the Google platform for, I don't know, now like 12 years, something like that, 13 year long time. But every week, we're going to be able to do that.

Because Google was working, rightly so, to support schools, things would change. So we'd have to send information out to teachers and tell them, before it would look like this. Now it looks like this. Here's how you access your Google Classroom. Here's a new thing you can do in Google Meets with your kids. And just really had a group of people that worked really hard to stay on top of things and try and support teachers and students.

Seth Fleischauer (04:58.446)
Hmm. And so is that like the primary role of someone who's running IT for district? You're essentially supporting teachers with their use of their adoption of technology. Do you also have a role in the in selecting student facing materials? Are you overseeing the whole thing? Like, like, what exactly do you do?

Leigh Anne Scherer (05:20.558)
Yeah, so I have a variety of different roles. So before I moved into technology, I actually was worked, I was a English learner, teacher of English learners at the elementary level, and then I was a principal, and then I was a program coordinator. So I have that instructional background is actually more of where I've spent my career. And so it varies from district to district. In my district,

I'm not on the adoption team, so to speak, so I'm not choosing the curriculum, but I advise and I do have some coaches, some teachers that work for me that actually sit with the other teachers as they're evaluating programs and make recommendations. A lot of what we do in technology when it comes to digital curriculum is more making sure that it will function within our ecosystem and that

you know, that we can roster the students so that when they log in, they see what they're supposed to see and the teachers can see the right students. But a big piece, I'd say, of my responsibility when it comes to digital curriculum is really sort of that security piece around data privacy for students. And so that's the particular lens that I think I bring to our district's adoption process is.

ensuring that the practices of the companies that were the vendors that we're working with, that they're in line with what our data security requirements would be.

Seth Fleischauer (06:52.558)
Got it. And so that's the primary thing you're looking at when you're evaluating programs is that lens of data privacy. Are there other things that you're looking at in terms of what you would recommend that people put into the classroom versus what they shouldn't?

Leigh Anne Scherer (07:09.23)
Well, I have my own beliefs when I see something. If I think it's a good product or not, in my particular role, that's not really my job to say, well, I think that's a dumb math program. But I would say one of the pieces is accessibility. What sort of accessibility supports are in there for students? So are there language translation tools? What's available to be read aloud to the student? Is there text to speech?

is their speech to text, is all of those, how can they adjust the color? So a lot of that accessibility is something that if I'm a math specialist or I'm a reading specialist and I'm evaluating a program, I might not necessarily have that lens to look for those things. And so that's definitely something that we've tried to make much more explicit in our adoption process when it comes to digital curriculum. Because I might think that,

because it's on the computer, all these things will work. Well, we, for example, just one example, we're, you know, days away from signing a major contract because with a, with a, with a company and the teachers on the team were really excited about it. But the, the student facing piece would only work on iPads and we have Chromebooks. And so we could have gone down a really bad path. yeah.

Seth Fleischauer (08:31.638)
Hmm.

Leigh Anne Scherer (08:38.222)
Exactly. So there's a lot to be aware of, a lot more than I knew before July of 2020. We had to keep an eye on. So it's been a learning curve, but our district has a great team. So it's been good.

Seth Fleischauer (08:45.902)
Yeah, yeah.

Seth Fleischauer (08:52.526)
Awesome. Yeah, and it's interesting that you talk about accessibility. I'm thinking back to the presentation that you made at the University of Portland on AI policy. There was a man in that room with us named Bruce Alter. He's down in Tigard, I think. And he was talking about how important it is for us to consider accessibility when we're talking about AI tools, arguing that

Students with disabilities, especially are some of the people who have the most to gain from using these tools, which I thought was a really interesting perspective. And I think that allows me to segue a little bit into the framework that you've developed in your district for AI, which I understand it as essentially a framework for experimentation where the you know, it's like these are the things that we should consider. And let's go off and have, you know,

these little experiments running each, you know, one teacher at a time. And then at a later date, we're going to come together and sort of pull all that data, all those experiences into something that's maybe a little bit more formal and directive. Whereas right now it's just sort of supportive and experimental. A, do I have that right? And B, could you give us a little bit more information about this framework that you've provided them?

Leigh Anne Scherer (10:16.046)
Yeah, yeah, that's you. You stated it really well. I think at this point, we don't feel ready to make any hard and fast rules. And part of that is because, you know, the technology is changing so rapidly that we could make a decision today and next week it could be irrelevant or it could be outdated or it could have the opposite effect of what we think whatever that decision is. So.

Seth Fleischauer (10:34.286)
next meeting.

Leigh Anne Scherer (10:44.27)
I started, along with the rest of the world, sort of started hearing, I guess, the rest of the world that aren't technology AI geeks, right? Started hearing about this AI thing in November of 2022, right? And then I had, because I'm fairly new to educational technology, I had set myself up for a lot of learning experiences that year. So I was going to several conferences, I was meeting with some leadership teams, things like that.

everything after that was all AI all day long. So I felt like I just sort of got dumped in the deep end with sort of all this learning of all these new things because the world changed, you know, sort of overnight in the educational world along with it. So this was all happening. And at the same time, I felt this, you know, it's really big, it's really important, but that wasn't necessarily the experience of, of.

other folks in our organization because they hadn't been hearing it in the same way that I had. So we started having discussions, we started talking about it, and we started hearing from teachers. And that was really what motivated us is we were hearing from teachers and teachers were wanting, they were wanting some guide rails and they were wanting some, you know,

Some teachers wanted us to block everything and not let students use any AI. Other teachers wanted everything wide open and not to have any blocking and not to just to leave it completely up to students and teachers to determine how they were going to navigate this new technology. And we needed to come somewhere in the middle. So because we really weren't ready to make any sort of...

you should do this, you shouldn't do that, or you can't do this, and you can do that. We did come up with this idea of experimentation. And so as our district has never blocked anything for outside of our normal security protocols that we have in place, we haven't blocked all AI or opened up specific things even. So we're following our regular process where...

Seth Fleischauer (12:41.422)
Hmm.

Leigh Anne Scherer (12:57.998)
If a teacher wants to access a particular tool, there's a review process. And so they request to access a tool. If there's something out there like ChatGPT, for example, we allowed it to, we allowed adults to log in with their Gmail account, with their North Clackamas credentials. So we've pretty much left it open to adults and we follow our...

our normal adoption process for students, which would require a review for security, for appropriateness if it matches with our curriculum, things like that. So we haven't like outright blocked anything, but we haven't promoted particular tools. And the other thing is we don't know what the funding model is gonna look like or what the pricing models are gonna look like, because there's such a range. I think some are looking at like $30 a month per user.

And some are, another one I was looking at is $4 .50 a year per user. So, you know, that's a pretty big range when we're talking about 20 ,000 users for our district. Right, me too. Exactly. So, the other thing is we didn't want to adopt something if, because there's lots of deals, you know, you'll get a deal and they're going to even give you something for free for six months. We didn't want to do that and then get our teachers and students used to using a particular tool.

Seth Fleischauer (14:02.83)
Eh.

Seth Fleischauer (14:07.662)
Yeah, I'll take the 450.

Leigh Anne Scherer (14:27.662)
and then have to say, I'm sorry, we can't afford that. So we haven't adopted anything as a district. And so it really has been driven by teachers. And we have a small group that's been meeting regularly throughout the year just to talk about what's out there, what are they finding in their classrooms, what are some ideas for practices that we might think about as a district. So that went this school year, the 23 -24 school year.

And then I think we'll probably revisit that, I would imagine, for next year, but we haven't had that discussion yet. So we're just wrapping up the school year now.

Seth Fleischauer (15:05.646)
Awesome, okay, I have lots of questions. One of them is that you started in talking about your lens of privacy. Then you're saying that you're allowing teachers to log in to ChatGPT using their Clackamas emails. I'm wondering what convinced you that that was safe enough for them to do?

Leigh Anne Scherer (15:13.806)
Yeah.

Leigh Anne Scherer (15:22.798)
Mm -hmm.

Seth Fleischauer (15:30.67)
And like what guardrails are in place for protecting, for example, student data that a teacher might be like, here's my roster with these results, like grade my kids for me. How are you ensuring that that initial lens is considered during these experiments?

Leigh Anne Scherer (15:49.87)
Yeah, so I don't know that I'm convinced yet that it's safe. I'm trusting our teachers. We've done a lot of communication with teachers of, you know, you don't have to use any AI tool, and we're not telling you to not use a tool or use this one. If you do, here's some things to think about. And one of those that we've stressed over and over and over again is the need to not put any personally identifiable information in a chat tool, right? So, which is also why we haven't opened up

Seth Fleischauer (15:52.494)
Hahaha!

Leigh Anne Scherer (16:19.758)
any AI enabled tools other than, let's see, where a student is going to actually interact with an AI. We haven't enabled anything yet because we don't know what the kids are going to put into the AI tool, right? So that's definitely a higher level of risk. But we've done it through communication, through repeated reminders of the importance of protecting student information.

And then just examples of how you could ask a question or how you could get information. But there really isn't any way to keep it completely safe. But we trust our teachers with this in other ways as well with this information. So that's really in like what we're trying to do is keep it in line with our other practices. And the only way we'd be able to really 100 % insure is if we completely blocked any

Seth Fleischauer (17:08.046)
Hmm.

Seth Fleischauer (17:13.23)
Hmm.

Leigh Anne Scherer (17:19.246)
AI enabled and even then features would be able to theoretically go on their personal computer on their personal network and then we'd be in the same situation. So.

Seth Fleischauer (17:25.998)
Yeah.

Seth Fleischauer (17:30.062)
And so you said that you haven't approved any student facing AI tools, but you also haven't banned CHEP GPT. So is there an assumption that students are using this to a certain extent?

Leigh Anne Scherer (17:34.766)
That's it.

Leigh Anne Scherer (17:43.726)
Student filters are a little different. So with our, yeah, I'm assuming students are 100%. I 100 % assume they are. Now, are they using it with their North Clackamas email account? That's what I can control. I can't control a student doing this at home. Now, and I'll have to remember how we have this set up because actually for our high school students, they may have access to certain...

Seth Fleischauer (17:47.086)
Okay. Okay.

Leigh Anne Scherer (18:12.014)
they may have access to chat GPT because I think now it's at age 14.

Seth Fleischauer (18:16.526)
Yeah, 13 to 18, they can get parental permission and then they can get an account, I believe.

Leigh Anne Scherer (18:21.582)
Yeah. Yeah. Yeah. So maybe we have it. I'd have to check. I don't think that we've enabled or that we've unlocked that. In our filters, we can filter out anything that has sort of that AI just on the internet. If they're logged into their North Clackamas device and or if they're logged into their personal device on our network through Chrome with their North Clackamas credentials, they wouldn't be able to access that.

Now that said, this summer we are doing a pilot with an elementary reading, AI enabled reading sort of tutor, coach curriculum or program. So that's our first one that we're trying out.

Seth Fleischauer (19:03.15)
Got it. Yeah. One of the things that I think about when I think about student access, even though it feels as if we're not quite ready or at least maybe the tools aren't quite ready, is that equity piece, right? Where students, you know, so if you're blocking it with their North Clackamas account, that's, you know, but students are going home and using it. What about the students for whom their only email is a North Clackamas account, right? So I assume that is that part of the motivation in seeking out a student facing tool that can work?

Leigh Anne Scherer (19:24.31)
Yeah. Yeah. Thanks.

Leigh Anne Scherer (19:31.598)
Absolutely, yeah, absolutely. And definitely, and I think your earlier point about which students are going to benefit the most from these tools are the students that this sort of enables a greater level of accessibility. So if a student is not able to speak fluent English yet, right?

Seth Fleischauer (19:46.382)
Mm -hmm.

Leigh Anne Scherer (19:57.646)
they can get some coaching and feedback on their written responses to better capture what they're trying to say. Same thing for a student, to be fair, with a language processing disorder or something like that, that ChatGPT is a great tool. And I'm using ChatGPT, but Claude or any of the other sort of large language model chatbot type tools can be really powerful supports for them.

Seth Fleischauer (20:24.75)
Mm -hmm.

Leigh Anne Scherer (20:25.55)
So I think that's where we're sort of in that experimentation. And we do have a few classes. I'm trying to think. I think we have a few teachers that are in sort of our, I want to call it a focus group, except it's not really set up like a focus group, more like a pilot teachers, we're calling them, but they're not really pilot teachers anyway, that we've sort of opened up broader access. So for example.

Seth Fleischauer (20:53.454)
Hmm.

Leigh Anne Scherer (20:54.35)
I mentioned we're a Google district. So there's the free version of Gemini. And so we've opened that up to a set group of folks who we feel really confident that they understand the need for data privacy and security to see how that might function.

But others, it would just be if they don't need to log in with their North Clackamas credentials that they would have access. And we're not blocking it on our network.

Seth Fleischauer (21:26.446)
Yeah. Okay. So I'm hearing there's a couple of layers of experimentation here and you did mention language learners, which is something I'd like to get into right after my next question. But you have these these layers of experimentation where some people have greater access than other people, but there's a general access that's being granted to all of the teachers.

You've got a group of people that you're meeting with regularly, at least in this school year, and you're getting, my question is about the type of feedback, the type of data you're looking for from them. Like, are there specific things that you're measuring quantitatively, or are you just looking to get stories and try to kind of piece together from there?

Leigh Anne Scherer (22:06.67)
Yes, definitely in the storytelling phase. It's definitely in the, hey, I tried this and it worked great, or, ooh, here's what I noticed. Here's what I used in my planning. Here's what I used with my students. Here's how I set up a lesson. It's that sort of thing. And a lot of it is even not what I'm doing, but how I'm thinking and feeling about it, right? So to get a sense of how comfortable am I as a professional teacher.

Seth Fleischauer (22:30.478)
Hmm.

Leigh Anne Scherer (22:36.27)
in the classroom feeling with this new technology, what do I feel like is the line? Right? And so it's really, it's having those philosophical kind of discussions, not so much, hey, I found this great tool and here's how I used it. There's some of that, but a lot of it is more sort of philosophical and where are we going to fall as a district? You know?

Seth Fleischauer (22:46.19)
Yeah.

Seth Fleischauer (22:56.526)
Yeah. Are you finding that people are down for that conversation? Because I love talking about that kind of stuff. I find that like some people will like play with me and others won't.

Leigh Anne Scherer (23:02.382)
Young.

Leigh Anne Scherer (23:10.222)
I think it's this particular group are sort of the ones who wanted to talk about sort of policy or accepted practices. So I think they're open to that because that's where they were at. So they were drawn, you know, they were asking the question, which is why they were invited to participate basically.

Seth Fleischauer (23:28.75)
Yeah, yeah, makes sense. OK, so you talked about language learners and AI. One of the things that intrigued me about our previous conversation is that you said that there are a lot of parallels between how these large language models learn language and how we are how we ourselves learn language. I've often said that that we are the world's best GPT's. But can you unpack that a little bit? Like what are what are the parallels between how these machines work and how are learn and how we learn?

Leigh Anne Scherer (23:58.99)
Well, and I will say this is my understanding. So if someone's listening to this and they've trained a language model and they say, no, that's not how it works. And just preface it with that. So I, yeah, please. so I, you know, just have, just in my reading, just on, you know, I, it's, you know, I don't have a lot of time, but just in the things I was reading this one particular article, and then I was, I'm reading a book now that reinforced this. So I feel a little better about saying this is how it works. Cause this is my understanding. let me grab it. Hold on.

Seth Fleischauer (24:01.71)
Hehehe

Seth Fleischauer (24:07.566)
Yeah, please let me know so I can update. Yeah, thanks.

Seth Fleischauer (24:26.094)
What's the book?

Leigh Anne Scherer (24:34.382)
It's called Co -Intelligence by Ethan Molyk. It's very good. Look at all those tabs. Look at that.

Seth Fleischauer (24:35.662)
Hmm cool, I'll put that in the show notes.

Leigh Anne Scherer (24:43.822)
Yeah. So the way language works is you build your vocabulary by connecting known words to other concepts and other words, right? And so they're linked sort of in this network of vocabulary. And then there's all the grammatical pieces in all of that, right? But those are rule -based. And so I would imagine that some of that is programmed in...

Seth Fleischauer (24:58.03)
Mm -hmm.

Leigh Anne Scherer (25:10.286)
But then the way you sort of internalize that as a real human being, because you can learn a rule as a real human being, but you're not going to be able to apply that until you're really fluent in the language and you sort of instinctively or without thinking apply that rule. But the model works the same way. And as I was reading it, it was really fascinating to me because it mirrored to me language acquisition, right? Where you're learning conceptually,

about something and then you're linking your current knowledge with new words and new concepts and new ideas. And so then it just becomes this network and that's how it grows. And so with the language model, with the large language models, they're fed certain things and then they're given feedback that, you know, this concept of car is related to this concept of truck. But there's, you know, wheel is related, but it's not directly related in the same way. Right. And so.

Seth Fleischauer (26:08.046)
Hmm.

Leigh Anne Scherer (26:09.55)
you sort of start thinking about these networks of words and you have this both morphology of language where like if we think about in English, you know, our Latin and Greek, the root words and the base words and the prefixes and suffixes. And so if they share that, they're related by meaning. So you then can gain that network of understanding of the word because you know the parts and what the parts mean. And you know that

biology is related to biome because it has that bio, right? Or biology is related to geology, but it's because of the ology. It's the study of. So if you know those word parts, then you can begin to understand the meaning of the word. And so as I understand it, the language model is sort of built the same way, that it is sort of adding to and then creating a network out of

Seth Fleischauer (27:03.694)
and then reading the different nouns. And the other thing that I was reading out was this.

Leigh Anne Scherer (27:07.566)
new words. And the other piece that I was reading about in this, I think it was in this Co -Intelligence book, was about the idea of getting feedback and that the best models, LLMs, are getting sort of a human interseeding in the feedback and, yeah, that's right and that's wrong, and that it's growing its understanding based on that human feedback. Well, that's actually how we learn language too, right?

Seth Fleischauer (27:23.734)
Mm -hmm.

Leigh Anne Scherer (27:32.974)
So if I'm interacting with you and I say something and you give me a look like I don't know what you meant, that's feedback to me that I didn't phrase it correctly or I need to restate what I thought. So it just was sort of eye -opening to me with what I know about how people learn language, that it's, and I guess, you know, duh, it's a language model. It would, you know, it would make sense. And it just, it did seem to be those parallels.

Seth Fleischauer (27:54.222)
I'm sorry.

Seth Fleischauer (28:01.134)
Yeah, so, because like I think about how even now I'm thinking way too hard about how the words are coming into my brain right now. But it starts with like an idea and then I begin talking and there are rules that dictate which kinds of words can come after each word there. And then I have this idea, the seed of a thing that I'm trying to put into.

Leigh Anne Scherer (28:11.31)
Thank you.

Leigh Anne Scherer (28:24.91)
Nothing.

Seth Fleischauer (28:30.894)
these like little blocks that I can arrange into a line that mean that thing that I'm trying to say, right? Is it is that your understanding of how the language models work? Because or does it does it lack? Maybe it's getting too philosophical. Does it lack the idea? Right. And it's just predicting the word, right? Because that's my understanding.

Leigh Anne Scherer (28:51.022)
I think that's the question, right? I think that's the question that we don't know the answer to. Because at least what I've seen in my tens of hours of thinking about this or looking at this, is that the people who are doing the eye don't know exactly why it works. Right? I know. But it's that...

Seth Fleischauer (28:56.462)
Yeah.

Seth Fleischauer (29:09.358)
So creepy. Yeah.

Leigh Anne Scherer (29:14.958)
Yes, there's pattern recognition, but pattern recognition you could explain, right? And you can't pattern recognition unique ideas and unique content. And so that's where, to me, I see that really deep connection with language because we have theories of language acquisition, but we don't have here's exactly how it works and here's why. Right. So how do you get from that magic? Exactly. Right.

Seth Fleischauer (29:37.486)
Hmm.

So we're creepy too.

Leigh Anne Scherer (29:44.334)
But that's why I think it's, that's why to me it seems like the same thing in a lot of ways. But yeah, I think the idea, the concept, like you can predict which word is most likely to follow another word, which is part of the algorithm, right? It's part of how it works. But, you know, I'm looking at something right now as the, so,

Seth Fleischauer (29:51.118)
Yeah.

Leigh Anne Scherer (30:12.238)
The is going to follow as a certain percentage of time in the whole realm of what I've looked at to compare it to. But how am I going to know that that's appropriate as the, and in other situations it's not? Because you would just get a bunch of gibberish. So there's that feedback loop of this is a real thing, this is a real idea. And that's, I think, where that human interceding or giving feedback can help.

which is why when they're training a language model, it's really important, it's really helpful that lots of people are in there, which to bring it all the way back is like my concern about students, right? So having our students giving that feedback and having their private information out, feeding something that we don't know what's gonna happen to that private information, and the student could be sharing really personal, specific information. And so that's why I feel a much greater level of

Seth Fleischauer (30:39.022)
Yeah.

Seth Fleischauer (30:49.742)
Hmm.

Seth Fleischauer (31:05.39)
you

Leigh Anne Scherer (31:08.91)
concern about student access than I do the adults in our system.

Seth Fleischauer (31:13.326)
Yeah, and I think that that's pretty universally the baseline requirement for student facing AI tools is to have it be a walled garden where you're not sending stuff back because you can trust your teachers all you want to not share student data, but I'm not sure that you can train a lot of kids to where the line is, right? Maybe when they're a little bit older. It's interesting hearing you talking about unique ideas.

Leigh Anne Scherer (31:22.414)
next night.

Seth Fleischauer (31:41.902)
cause one of the things that I run into sometimes when I personally use AI is how bad it is at marketing, right? Like I try to get it to like, write me a good headline for like this thing. And like, it just can't do it because it's trained on all of this stuff that already exists. And part of marketing is like, that's a new and interesting way to think about a thing. Right. Whereas you also hear about the hallucinations that happen. Right. And so, and then I'm thinking like,

are good marketing ideas, simply hallucinations with like some rules built in so that they actually work. And so it's like, why is it like, is it just that there's a special instance of the use of the tool where you're actually going to get something new and innovative? It just tends to like happen randomly or happen with a good prompt or is...

Leigh Anne Scherer (32:16.846)
Yes!

Leigh Anne Scherer (32:37.39)
I mean, I think it does get into it's how you have the conversation, right? And that was one of the things I, this is very accessible. I've recommended it to lots of people now. But it's like there's prompt engineering, right? Or like patterns, but it's sort of like setting the stage. And so what he was talking about is this idea of you start with,

Imagine you're this, you're a marketing professional and you are really innovative and you have new ideas that you want to share. Now what I want you to do is give me 10 ideas for, hmm, right? But you set up the premise from the beginning that it's innovative, not, okay, here's tried and true. And that I haven't tried this myself, but yeah, that's what I was reading about. But that gives you better results.

Seth Fleischauer (33:27.182)
Yeah.

Seth Fleischauer (33:31.75)
All right, yeah, school. All right, all right, schooling me on my prompt engineering. I like it. I'll try. I'll try that and report back. I did also want to get into language learners themselves because I know that your background is in that. You know, when we talk about the potential for students to use AI tools in my work, working with language learners, we have some special considerations that I think we're making for them, making sure that they're not bypassing the cognitive.

Leigh Anne Scherer (33:45.486)
Yes.

Seth Fleischauer (34:00.494)
processes that are necessary for language acquisition by using these tools. How do you think about that special population of language learners within this conversation of especially student facing AI tools?

Leigh Anne Scherer (34:16.398)
I think there's two things at least that have been at the top of my mind about English learners. One is teachers using AI as a tool to reach students, and then students using AI as a tool to more closely mirror native English speaking, right? Just in the context of US school system, like I feel like I have to preface all of this because of my language background, right? But.

US schools teaching English to students whose home languages are other than English. And from a teacher side, one of the things that I've been very cautionary because of my language background, my language teaching background, we have a lot of teachers who really want their students to access the content, which is what we should. Right. I'm a history teacher. I have a textbook. I have all this information.

Seth Fleischauer (34:49.806)
Yeah, thank you.

Leigh Anne Scherer (35:14.094)
I want all of my students to equally be able to access. Now, AI is a tool that can help with that because the translation tools and the taking of text and sort of reframing it in a way or summarizing bullet pointing, all of those types of ways that we might use AI are powerful, helpful ways for students that can help students better access the core content. And...

If we are just translating everything, right? I had a teacher who was requesting access to a particular tool that would go in and translate all of her slides into multiple languages, which is great, but she's not doing anything else to give students access to the content and to develop their English language at the same time. So if we're relying on translation,

That is only half of the picture of what we're trying to do for our students who are developing English, our multilingual learners, is they are learning English and content at the same time, right? And if we're just translating, all they're doing is learning content, they're not learning English. And so I think there's a real, I actually think there's a danger there. That's one of the things that, besides student privacy, that I'm most concerned about is because it's powerful and good.

and easy, and it's going to leave out this whole other part of how you teach a student when they are not a native English speaker or they aren't fluent in English in my classroom. So that piece is, I will say that's the piece I'm most troubled about because the benefits are so great for the student, but it's a short -term benefit, not a long -term benefit. And...

So anyway, so that's one of the soap boxes that I get on sometimes when a teacher will ask for something or I'm working with my team, the teacher team I work with, they get very excited about some of these tools. And I'm like, yeah, but remember, we're teaching English and so we don't want to just translate everything and then have the tool read it aloud to the student if they're not actually literate in that language, right? Because that's another piece, developing literacy in the first language.

Seth Fleischauer (37:33.198)
Yeah.

Leigh Anne Scherer (37:35.31)
And then how do we teach students, let's see, so students need some sort of language base if they're interacting with the tool in English, right? So depending on what tool we buy or what tool is out there for a student, it may or may not be trained in other languages. So there's accessibility for that as well. And then how do we teach students, all students, but particularly English learners to use this as a tool?

tool to develop our own writing skills, our own, you know, thinking instead of just doing it for me. And I think that's a piece that I don't know that any of us have the answer to that, but I think we need to continue thinking about. Something that just occurred to me too is I heard someone talking on some show about,

you know, if you're in like a tutoring tool, like an AI powered tutoring tool, and it's going to give you practice sets, or it's going to, you know, it's going to give you, it's going to take you there. So I'm going to interact with this tool. I'm going to give it, you know, respond to some prompts, and I'm going to give it some information. And then the tool is going to determine, okay, here's the next step, here's the next step, here's the next step. And we're going to have the potential to,

Seth Fleischauer (38:55.022)
Yeah.

Leigh Anne Scherer (39:01.838)
continue some of the biases that are built into that and continue students sort of on a trajectory that's more flat instead of a higher trajectory because the AI is only going to be able to know here's the step and here's the next step, not how do I maybe accelerate. And I think the same problem could be in place for our English learners who are interacting with tools that are sort of built to teach something, you know?

to sort of take the place of the teacher, that if the tool doesn't have all that other concept about the student, that they're three years behind and they should be here and we only have six months left and here's all these other pieces, it's not smart enough because it's not the teacher that's sitting in the classroom with the student and knows the student. And so it's how do we work together with the tool so that we can really make the most of the personalization and the...

the differentiation that's really difficult to do, but an AI can do it, but we have to be the ones to help guide that and to make sure it's not going off track.

Seth Fleischauer (40:02.126)
Yeah.

Seth Fleischauer (40:09.934)
Yeah. And I think there are some tools out there. I know some that you and I have both played with a little bit where you can give it that context. And you can program individual tutors for individual students, provide them with the context about the student without violating privacy. You can also program the AI to be a tutor. So that's a good way to do it.

Programming it as a tutor gets you past that bypassing cognition problem. Whereas personalizing the tutor with information about the kid gets you past that flat growth problem. And so I think that if you combine those two, I mean, that's the ideal, is that we can use that so that then the students are using the tools, they're using it to develop their language.

Leigh Anne Scherer (40:54.83)
Yes. Yes.

Seth Fleischauer (41:06.03)
By going through these steps, but they're also doing it in a way that's going to make sure that it's dynamic and not static So that's the dream and maybe by the time this episode comes out that dream will be a reality. Yeah

Leigh Anne Scherer (41:17.294)
saying, I'm next week. And I think, I think the having been in education long enough to know that we look for a quick fix, right? And so we're going to be sold. We educators are going to be sold. this tool, all you have to do is the kid sits down at the computer and they start interacting and the tool does everything for you. Right. And there there's it's good enough that it fools us.

Seth Fleischauer (41:38.67)
Hmm.

Leigh Anne Scherer (41:45.07)
into thinking that that's what it's doing. But that's where we really need to be aware of the danger of giving too much power to something that we don't really understand and doesn't have the student's best interest at heart, right? A company is trying to sell a product and they do want to help kids and it's a machine, it's not a person.

Seth Fleischauer (42:17.07)
Amen. Well, that is that's a great place to end. Is there any place on the Internet that you would like our listeners to find your work? You're like, I'm in the public sector. I just do my job. Excellent, William. Thank you.

Leigh Anne Scherer (42:26.702)
no.

Leigh Anne Scherer (42:33.742)
about right.

I mean, I want interest, you know, but different type of work.

Seth Fleischauer (42:41.294)
Well, thank you so much for your time, for your candor today. I really appreciate you joining us. Thank you so much to our listeners for tuning in. If you'd like to support the podcast, please do tell a friend, leave us a rating, a review, or follow us. Thank you as always to our editor, Lucas Salazar, and my advisor, Deirdre Marlowe. And as always, if you want to bring positive change to education, we must first make it mindful. See you next time.