Ep #13: AI Invited - Responsible AI Use In Schools Without Offloading Learning

We dig into what AI is actually solving in schools and why “just because you can” is a bad rule for classroom tech. We share real classroom moments, practical guardrails, and the skills students need so AI supports learning instead of replacing thinking. 

• AI as a tool that reveals weak assessments rather than “creating” cheating 
• research projects and why Copilot is not a source 
• credible sources, verification, and the problem of fake citations 
• Magic School sandboxes that ease cognitive load without giving answers 
• transparency and academic integrity when teachers and students use AI 
• banning AI versus the calculator comparison and why outright bans fail 
• student privacy, contracts, and deliberate rollout over rushing adoption 
• AI literacy as critical media literacy including bias awareness and emotional manipulation 
• real workflow example using multiple AI tools
• guardrails as shared values and consistent expectations across teachers 

Reach out, message us on the Facebook, the Instagram, shoot us an email. We love the fan-mail and thought about next season's topics! 

[Transparency Note: This full episode description was created via CoHost AI through Buzzsprout, proofread by a human]

Transcript

Welcome And Why AI Matters

Ben Webb 0:09

Welcome back to the pod. This week, artificial intelligence. Just because you can do something doesn't mean you should do something. So we get to talk about AI, its implications in the classroom, what we should and should not be doing, and how is it changing the role of a teacher? Stick around. My name is Ben Webb. This is Andy Govey. This is Live in the Lab Schools.

unknown 0:35

Mr.

Ben Webb 0:35

Govey, we are to the topic that I've been wanting to talk about since the beginning.

Andy Goveia 0:40

This was in the the so for everybody listening, we've we've been sharing a notion planning notebook online, basically. And one of the first when we started this idea back in August, one of the first episode ideas was AI. And it's been sitting there because we prioritized people. As you should. As you should, which is very fitting that now, after we've prioritized people, we'll now talk about robots.

Ben Webb 1:11

And just thinking back to the beginning of the school year, how so much has changed in such a little amount of time. And it's going to be completely different if you come back to this episode in next school year. Yeah. And so I'm excited about this one. I think this year, especially for me, it's been a big exploration of what we should be doing in schools in regards to artificial intelligence, how it has changed from okay, gen uh gen AI to agenic AI to we're just trying to understand what an LLM does. And so for me, I'm just excited to kind of see am I even in the right wheelhouse? Am I, as a colleague, like, are you seeing the same things that I'm seeing in the classroom? But also, like, what are the things that we should be looking ahead to? So a little bit of a futurist kind of a perspective, but also just the

What Problem AI Solves

Ben Webb 2:05

challenge of it's it can do so much. Yeah. So I guess first question I want to throw to you what problem is AI actually solving in schools? I think when we talk about these are tools, and I think one of the hot takes, I so full disclosure, using perplexity AI, using Magic School, using something like the Notion AI that we have, I asked for some, what are some hot takes?

SPEAKER_01 2:35

Yeah.

Ben Webb 2:36

The first hot take that it gave me was AI isn't making students cheat more. It's exposing that our assessments were already easy to cheat on. And I thought that was such a wonderful quotation to get from an AI. And then the first question, of course, was okay, what is AI actually solving? So choose your own adventure on this one. How do you feel about that?

Andy Goveia 2:58

I that hot take is just it's the here's the problem. It's such a hot take because pedagogically, students are able to cheat when there's a flaw in the system you put in place. Oh, agreed. And so just because AI can do a task doesn't mean that your assessment was easy to cheat on. The assessment wasn't broken. It wasn't broken. It's just that this tool is able to get around the what you've may have created. Now, I'm also not defending the fact that, yeah, some students do cheat. It is a reality of what we run through in in teaching, and you have to figure out how to handle it, right? Like here we are right now, and we're doing a research project. I think you all are as well, doing a research project in my class right now. And I have a couple kids that somehow, it's like somehow Palpatine returned, somehow got into copiloting. Um, and we've been talking through like they're citing copilot as their source. This is the maybe this is the brilliance of why I disagree with it's making easier to cheat. Kids are just saying they're using it. Like in before they'd go, no, I I didn't, I know. What do you mean I was using AI? Now it's the cited source for a couple of these kids. And once we're sitting down going through their outlines that they've done in the research, I'm like, copilot's not a source. Shouldn't have used it. It's not a source. Well, let me go see what sources it used. And I said, that's why I don't want you using co-pilot magic school chat. I don't want you using these AIs for content acquisition or content learning because it's pulling from somewhere else anyway. YAT neatly packages it for you. You need to go make sure that source is credible because sometimes the source that it's going to spit out is not credible at all.

Ben Webb 4:38

Right. It's it's a it's a fiction, it's not even a real source. It just happens to look like it has an APA citation. Right. And I shared this during the Lightning Launch AI symposium we went to at ISU. Yeah. Shout out to the College of Ed. That was great. Yeah, it was a wonderful day. And one of the things that I shared out was like, I am curating some things in Magic School for my kids. And I've got kids who are asking me, hey, can you also make something that does this? And I get to have that conversation about how that tool I don't want to make. I don't want that to be in existence. Yeah. So, like they, for an essay they did back during our Holocaust unit, they had a couple of different like prompts and a couple of different tools that I built in Magic School that were purposeful in a certain safe sandbox. Ask questions, review their work, give them suggestions. But it was always like, okay, you have to be the one making the and it's making the changes. The tool won't do that for you. I know Copilot could do that, but here's why I don't want you to use that. And I think I'm having those conversations right now in our research paper in the freshman level because they're saying, Hey, can you make a tool that does this? And I'm like, I want a human to be the one giving you that feedback. They're like, Can you make a can you make a a bot that would help me with my thesis? And I'm like, I could, but I don't want to because I want to be a part of that process as the teacher.

Andy Goveia 5:49

So it's really interesting. You and I took very different approaches to this. And so for their thesis with me, I didn't tell them that I had, so I built them a chat bot to help them. Ah I built one to help them, but they didn't get it

Cheating Claims And Real Sources

Andy Goveia 6:00

until after they had written one themselves.

Ben Webb 6:02

Okay, I like that.

Andy Goveia 6:03

So their job was, hey, answer your big essential question. I want it to be one sentence. And there were some kids that could not get it down to one sentence. They had three or four sentences. And so for those kids, I was like, hey, you'll see there's a magic school link in classroom. If you go click on it, it's gonna prompt you and give you some suggestions on how to make your answer more concise. It's gonna help you put the right words in place, but you're gonna input what you think the right answer is. And it's also gonna tell you if what you've written is unrelated to your question.

Ben Webb 6:34

Yeah, or even unrelated to the the rubric that I've provided. I've given the instructions to Magic School to like say check their work against this. Yeah. So I think, okay, so what I'm hearing is It's so interesting we took different approaches. I'm and I'm thankful we got to do the opposite thing because I I think that kind of leads to my second question, which is what is good use in a classroom look like? The metaphor that I always use is driver's education.

SPEAKER_01 6:58

Yeah.

Ben Webb 6:58

Like I understand that there is a tool that can do this for you. I understand that there are cars that can drive themselves and can navigate and give you good directions and GPS. But when it comes to the driver's test, you are the one that has to show mastery of those standards. So in the driver's test metaphor, the AI can't help you. Yeah. You have to show that you can do all those things. So does good AI use take that all away where it's like, yes, these are tools, but regardless, you can show that you can do everything without the tool even present. Yeah. Does that is that an agreement with you?

Andy Goveia 7:33

I, yeah. So like what I love your driver's ed metaphor. It doesn't work for my grade level. But the way I've tried to frame it for the kids, like when they're like, why can't I just ask a chatbot to do this or tell me the answer? I always frame it to them like, are you gonna just rely on this chatbot anytime you walk into any situation? Are you gonna always be looking at your phone, typing into your chatbot, what am I supposed to do to order my dinner? Like, you can't rely, you can't give up the cognitive load. Yes. Like that's no offloading. I don't believe you can cognitively offload to AI effectively in a classroom. Now, what I do believe is that AI can help you ease your cognitive load.

Ben Webb 8:11

I like that.

Andy Goveia 8:12

It can help like this is why, and I this is an unpaid endorsement of Magic School. Yeah.

Ben Webb 8:17

Um maybe it'll be a sponsor next year.

Andy Goveia 8:20

I'm an ambassador for Magic School. I did all their extra training. Um, and I love the product because you, as a teacher, put the tool in the sandbox for them that you believe supports what they're doing. So, like for an example of that, easing the cognitive load, for most of my summatives this year, I've given them a study bot that will point them to the resources we've studied. So if they go, for example, where can I learn about the camp system and the Holocaust? It's gonna spit out because it has access to my lesson plans or classroom posts, it's gonna spit out the day of the classroom post and the title of resources that they could go to to find those answers. It's not gonna tell them the answer, but it's gonna go, hey, it looks like in this reading on special trains, you could learn how they arrived at the camp system. It's gonna point them and help them ease the cognitive load of, hey, where do I find that? But it's not gonna ease the cognitive load of them having to make sure they know it.

Ben Webb 9:19

So the quotation that I want to give you, it's from an article about brain maxing that I was sharing with the English department. Brain maxing. Brain maxing. I I have gotten I am look at you, Jin Alpha, Z, whatever. Hey, I I did say in the last episode that I am uh different from introvert and extrovert. So so the the way that I would kind of describe it is you don't want their mind to be disincentivized. Yes. You cannot devalue the the fact that your mind and your thought process has to be the the determiner. You are the decider. Yeah. So I like I like that explanation that you were going with.

Andy Goveia 9:59

It's it helps ease the load, not replace or take over the load. Now, I will also say there are some tasks I've just given up to AI. Like there, like I, for example, I, and I think a lot of you all here do this too. I update an in set of information slides that are displayed on a TV in the hallway. And before I gave it over, I was editing a Canva every week at the tune of 10 to 15 minutes a week, editing the Canvas slides, making it look nice. And then, and again, I am fully, and I want to say this for everyone listening, I am aware of the data, the energy implications. I am aware of all of these pieces that affect that when you use AI, you're making a conscious choice to affect X, Y, Z. Yep. And I am well aware of the fact that when I do generate an image, one image a week, to handle the lunch menu, for example, or to handle our schedule. But I mean, that's gonna generate it in 30 seconds. I give it the text input, it generates, I have my image ready to go. That that load is perfectly fine in my case of I'm making something that is not original work, I'm not promoting it, I'm not publishing it. It's helping students know what's going on. And so it saved me now 30-ish minutes a week by offloading that specific cognitive load.

Ben Webb 11:22

One of the words we teach kids when we're talking about um bioethics is beneficence, maximizing the gains, minimizing the harms. Yeah. And so, like in that case, you are then offloaded to go do something else that's equally, if not more important, because you were able to offload this one part to

Good Use Without Offloading Thinking

Ben Webb 11:40

the tool. Yeah. Um I think that's really crucial. Like you're you're you're making that conscious effort and also you're being transparent about it. Yeah. One of the things that I heard a bunch um at a symposium in Champagne, Urbana was that's similar. Like you have to be transparent about its use. And I I mean, I put it comically at the end of all my emails that I'm, you know, using a bad Chat GPT or co-pilot in my case, like to proofread and check my emails and all that kind of stuff. But like you have to be clear, okay, this is the tool and what was its use, so that especially, you know, colleagues can then know, okay, this is what you did, but also for the kids that it's okay to use this tool. They shouldn't be afraid for use in certain cases, as opposed to like just like when the internet was invented, oh, I'm just downloading a paper and turning it into my own work.

SPEAKER_01 12:29

Yeah.

Ben Webb 12:30

There's a difference between blatant plagiarism and using the tool, but not offloading the tool to do the work for you and then saying that this is your own work. So, like, there's that academic integrity that you have to model as the teacher, I think, if you're going to use AI in the classroom. Yeah. Here's a second hot take that I want to throw at you. Yeah. Throw it. Because this is one I'm getting into argument arguments with my colleagues about right now. Banning AIs and AIs in schools right now is like banning calculators in the 80s. Technically possible, but educationally backwards. I have been waiting to ask that question of you all year.

Andy Goveia 13:06

So I'm I'm gonna actually go back to your FDA metaphor reference.

Ben Webb 13:13

Oh, yeah, before the before we started recording.

Andy Goveia 13:15

Before we started recording, we just were having a brief conversation about you wouldn't put something into your body that's not FDA approved. Yes, the tools are here, they're here to stay. But banning them outright, I don't think is gonna be sustainable because of simply what and I I don't believe that the workforce should dictate education. And I don't believe that education should dictate workforce.

Ben Webb 13:38

For those of you at home, there's just a uh look of pain on Andy's face.

Andy Goveia 13:41

I I don't, and I never believe that because, or I would say this even for middle school, I never believe that because of what high school needs to do, that middle school should be shaped by what high school needs to do, nor should elementary be shaped by middle, or vice versa. Just because what elementary does doesn't mean middle school should adapt, high school shouldn't necessarily adapt. There should be this fusion of what was, what is, and what will come next. So my very hot take on AI is banning it outright, terrible, not gonna work. Simply because what comes next in their life, competency in the tools may be vital to the job they're gonna do or the role they have in society, or just simply living. There are tools that can make your life so much easier at home that are AI and not what I think the the middle ground answer of what is, what was, and what will be. It's like the ghosts of Christmas past, present, future haunting us. Thank you, Dickens. Um or the Muppets, my personal favorite version. Yeah, as I say, that is the best version. Um you've got to just do it responsibly.

SPEAKER_01 14:44

Right.

Andy Goveia 14:44

You you can't, and this is I think our both of our schools have kind of done this a little bit. We really have taken it slow here in the lab schools on unfettered AI use. I mean, because I teach 13, 12, 13-year-olds, that magical SOPA law comes into effect on what kids can voluntarily agree to and not voluntarily agree to. We have a use contract with magic school. It took us until this year to get that use contract. So before that, it was just staff use only. Kids weren't really using these tools. And I'm grateful that I've been able to. We didn't ban it, we just took our time to get it right. And I think that's my answer to that hot take. Banning it, yeah. And and math class has adopted calculators and they've done it in a ped, at least when I was in school, a pedagogically appropriate way that it's a tool that supports not like we said with AI, it's not offloading necessarily, it's making the load easier. And I think to do that effectively, banning just isn't gonna work.

Ben Webb 15:45

Well, and personally. And I and I would agree with that. I mean, choose whatever metaphor you want for that. You if you make something off limits, that doesn't teach the kids why it's either dangerous, why it is important to understand the implications of that. Like to to go with just that whole concept. Like, I can show a kid where the line is in the in a classroom. I can show them when I pull up co-pilot in class or when I pull up my own Chat GBT or my own Gemini, which Gemini is blocked right now for that very reason. We're trying to do it deliberately and carefully and making sure that we are following SOPA and those kinds of like we have that responsibility as teachers to protect student privacy and data. And until we get that signed contract, we're gonna have to, you know, show kids this is what you can do outside of class, outside of like the school environment, but here's what you can do within that learning environment, and we're gonna kind of go over those literacy school, those literacy uh pieces. So we want you to be AI literate. I'm glad we don't have to teach them how to do certain pieces because with Agenic AI, you can kind of just give it the task and it will then find the right employees to then do the work for you. But I also think to just ban it outright, to ignore its existence is almost like we are neglecting our mission to prepare these students for their future. Whether that's going to be career-driven or not, I think right now in corporate America, we're seeing a lot of AI because we're being told to. But I also think we need to be the ones, going back to the Jurassic Met Jurassic Park metaphor that I always, or line that I always kind of cite, just because somebody can do something doesn't mean you should. Right. And I think we have to then, like the dinosaurs, evolve. And we have to see how our like the workforce is gonna change. Um, you know, how we then create and how we generate ideas and how we challenge ideas. Like all of that is really crucial in this process, but you can't rush into it and then say, oops. Yeah. Like three months into it, you can't be like, oh, well, all of this data is now out there, or we trained these models.

Andy Goveia 17:55

And especially with 12-year-olds like I teach, you can't and we're in a great system where like, yes, we could jump all the way in and go, oh, crud. But at the same time, what habit, what thing did maybe a kid take away that is gonna be detrimental for them in the future if you were to just jump in and do it that way. I actually think a really good comparative thing is the idea of one-to-one, one-to-one 15 years ago.

Bans, Calculators, And Student Privacy

Andy Goveia 18:20

Do you go one-to-one? Well, what'll that do if a kid has a a laptop all the time, or if they have an iPad all the time, what's that, what's that gonna do? They're just gonna be on it all the time, they're gonna be doing their work. And we as teachers figured out how to pedagogically to the best of our ability, make that work in our classrooms. I think AI is the same thing. You just have to figure out how to do it in a way that is authentic to what you're doing, that responsibly teaches what they can be ethically used for.

Ben Webb 18:54

I'm thankful that we are at the intersection of technology and education frequently. Yeah. And I think we're doing a lot of those things really well, but cautiously while also innovating and saying, okay, this is how I'm gonna graduate you out of the tools that I've made in magic school into now. You can make the tool. Now you can make the support, now you can create the thought partner. And that takes time. The same way that we want our students to take their time on a research paper. Yeah, I also want them to see that process that they go over to find your research, clarify the research question, create a thesis that captures the argument and then cite the evidence, then supports and proves their argument. Yeah. But that's what the AI is doing in milliseconds, but it's so crucial for them to be able to do that because the AI tool's not always gonna be there. Yeah. And I really, really, really want them to walk away with that mastery.

Andy Goveia 19:49

Yeah.

Ben Webb 19:52

For sure. Do we just end the episode there?

Andy Goveia 19:54

Is there more I mean, you tell me, you've got the questions. Well, okay, so I don't want to take I don't want to dominate, but at the same time. I this is a passion area for you. I mean, we've had episodes where I have been a dominating voice. I've listened to our episodes. I I can dominate the conversation from time to time. So having one where the Mr. Webb is featured, I'll make all my eighth graders listen to it so they know what they're getting into next year.

Ben Webb 20:18

Oh, I those poor eighth graders. Um welcome. Uh, so I think the next question that I would ask, and again, some of these are from perplexity, and some of these are me just kind of like vibe feeling out what you think about this. So, what skills do students now need because of AI? Like before AI, there were certain things we wanted them to be able to do. Yeah. And at times we want them to be able to use the calculator. At times we want them to be able to do and show their math and show all of their work. Show their thinking. Yeah. And sometimes when they write a paper, I do care about them turning in the drafts and saving the different versions of their work. But sometimes I just care about the final product. Because in English, we don't show our math, we show the final product.

Andy Goveia 21:04

Yeah.

Ben Webb 21:04

So, what are skills and what are things that students need now because of AI? How has it changed education? Because I mean, this is something that it didn't come from education. It didn't come from us saying this is the something that we need. If anything, it's us reacting to, well, now there is a nano banana too that can create an image of the same quality or less or more than what a student's able to do, what a teacher can do. Uh, an outline can be created by Gemini like that. And it could be accurate, it could be better than what I can make. It could also be worse. And if I don't check my work, it's gonna be probably pretty embarrassing in front of anybody that's there. So, like, how has it changed us? And I'll give you a couple of make maybe things you can go off of. Prompt literacy. I know you've been spending a lot of time in Claude right now. Yeah. Evaluation of output.

Andy Goveia 21:58

Yeah.

Ben Webb 22:00

Bias detection and originality.

Andy Goveia 22:03

That could be four things I would maybe throw out there. I think I'm gonna go with credibility and bias. I'm gonna throw them together a little bit. That I think now more than ever, and I'm gonna use people on Facebook. The amount of AI slop that I have in my Facebook feed sometimes because people share it believing it. Like there's some people on my feed legitimately sharing it, thinking it's real. Yep. Frustrating because take two seconds. No, ma'am, that pig is not really flying. Um so I think that's one skill that becomes even more relevant even across all the disciplines, is just that credibility of what you're seeing, reading, processing, and what bias or or maybe bias isn't the word. It's the ability to it's critical media literacy, is what it is and applying it to AI. Um, so if AI has created some sort of scholarly work, I put in quotes, what sources did it pull from? What is it saying? Is it accurate? Can you trust what it said? Can you verify what's in there? If it's something that is a visual image, is that image real? Can you verify that occurred? Where can you find it? Or does it look like a computer heard pig fly and it looks like a deformed pig with wings? On the flip side of that, if there is things being created, it's kind of getting into the whole when you talk about critical media literacy, right? Does it evoke a reaction from you? Is there an emotional charge it's trying to get out of you? So like the purpose of it. The purpose, right? Like why is this being done in this way? And so I think with AI being able to be so prolific in creation of things and outputs, students now more than ever, not that they didn't before, but it's

AI Literacy And Critical Media Skills

Andy Goveia 23:55

teaching them how to, even though this great tool and these outputs exist, hey, the critical and the critical analysis you still did on other things, it still applies. You can't, it's not an infallible product. And so that credibility, bias awareness, all of those things we would ask kids to do with any piece of media or creation double applies because you have to almost not you're not just checking a source, you're checking something that may be pulled from two or three half sources.

Ben Webb 24:22

So I'm gonna answer, I'm gonna, I'm gonna follow up with what you said. One of the questions I was gonna ask is what have you, what have you created recently or what have you generated recently? Because did you make that or did you prompt it to then co-collaborate and make something? So here's what I'm gonna do instead. What are some things that we've made recently that are cool? Because we've worked on a couple of things recently. Oh yeah. Um, but using a couple of different AI tools. So what are some of the things that you've made recently that are cool, are great, are really something you couldn't have done otherwise unless you spent like two full-time jobs doing it, but haven't yet passed the evaluation tool because we're not the experts to then give the stamp of approval on it yet.

Andy Goveia 25:07

Yeah. Okay.

Ben Webb 25:08

Because we've been writing bills. Oh, we've been we've been looking at school code. Yeah. We've made job descriptions. Yeah.

Andy Goveia 25:15

Yeah, we've been having fun. I'm I think I'm gonna use our our amendment we've written. Yeah. So we're going to Springfield in May to do some advocacy for the lab schools with funding. And my job was to look at evidence-based funding and come up with a proposal essentially to help us increase our funding. Now, full disclosure, no clue if it's going to go anywhere. We don't know if we're going to get any traction, but we're hopeful. We're trying. We're hopeful. We're hopeful. And so what I did is I researched EBF, I looked at funding distributions, I looked at our revenue, our expenses through the years, and I fed all of that into, I believe I started with Chat GPT, threw all of that in there and asked it to write me an amendment that would increase funding for the lab schools. And it spat out a 25-page thing. I then reprompted it a couple times to shrink it down. I realized I misprompted it from the beginning and I got it down to about a 10-page amendment. So then I took that and I threw it into Magic School for fun to go, hey, are you a legislative expert? Let's go. Magic School added some tweaks and adjustments that Chat GPT had missed. I then took that output, took it into Claude, and went back and forth with Claude a pretty extensive amount of time. Um it caught things neither of the other two had caught that I was also checking along the way, going back to the school code, reading legislation, going, Oh, yeah, okay, I get it. Here's where it came from. And so now we have a, I think, 12 to 14 page amendment that after a couple different AIs, and again, I took that Claude output and dropped it back into Chat GPT and Magic School and said, analyze, does this work? Everything I triangulated the triangulation of AI here, and it all said it should work. Now, in my disclosure, in the amendment, it says generated using these tools.

Ben Webb 27:21

Yes.

Andy Goveia 27:21

Um, transparency. It says generated by these tools, and we are currently sitting with that draft, that proposal, waiting for someone, whoever the powers that be, legislators, legislators, lawyers, someone who understands one to go, we think this passes a SNP test, it should work. Yeah. Or hey, propose it. They may make tweaks, but it does what you want it to do. Um, because fun fact, neither of us are experts in the school code. Surprisingly. Or legislators, or legislative policy people. Yeah. Right? We're not our thing. We are just lowly teachers who are kind of politics nerds. Yeah. Um, but terms of you asked the question, what have we made? That was a 25-minute dialogue with three AIs, and it spat out what I think is a passable amendment.

Ben Webb 28:13

Right. And I think so. The human element was not lost there. As much as we're talking about, like, okay, I threw it into this tool, I triangulated with these different sources, I took it and I threw it against uh open, I used uh ChatGPT and I use Copilot. I checked it through Perpex Perplexity and a couple of other um funding bills that I've created in the past to say, okay, would this also pass given the current environment of politics? So it's doing some additional research about what is what has and has not passed the General Assembly in Illinois. So like we did all of that, but it's still driven by humans. And the humans have to be the one who's either the lawyer who says that's gonna pass the test, the legislator who needs to be convinced this is worth the funding, because it's it comes down to okay, you're a human, you vote. And then also kind of like going over with all of the administrators. Uh, we brought the dean in into this as well. Like, so we can only create the pie in the sky. This is what we are hopeful for, and what we think will be good. It's the real humans and the real experts that then have to carry it over the finish line. You can only use the tool so far, and the tool can check its work, but you have to be the one to check the output. And so I think that's something where it's like, we've been working on that for a while. Did the AI able to process through a lot of data in seconds? Sure. Um, but some of those checks, it was a the it was a lot of conversation, it was a lot of back and forth. It was going through and saying to other people, associate deans, and saying, okay, yeah, that does pass the test, or that does make sense. Um, the bot was right. Um, and in getting that, it it takes somebody who's invested in politics, it takes somebody who's invested in, you know, going and advocating for their school system. It it goes back to us. Okay, we are the humans. This was our process, this is how the sausage gets made. But in the end, we think this is a good idea. Can we get this to happen? Yeah. So that I'm I'm glad it wasn't just, oh, I I photoshopped this thing using No, that that's the best, and I think that's the best example.

Andy Goveia 30:23

And we talk about easing the cognitive load, not replacing the cognitive load, right? Definitely I gave up a lot because I could have sat there and I could have written a not as polished amendment, right? That didn't speak to all the nuances of where we're mentioned in funding in the school code, which fun fact a lot of places, more than you would think. Um, and there are a ton of sections that needed a slight amendment to help with our funding. And so it the the beauty of this amendment, which is really a policy draft, is what it is, that contains amendment

Writing Policy With Multiple AI Tools

Andy Goveia 31:00

language, I should say. It's not just the amendment, it's an explainer doc. It's it's pretty thorough. It created what eventually was needed without my time being devoted to becoming an expert in writing legislation. And I think that's kind of the key to some of this is it alleviates the need to become an expert in certain things. But as you said, I think really well, it does not eliminate the need for the person to be knowledgeable about what it is they are prompting or they are asking or doing. I knew, for example, we are tier one for evidence-based funding still, which changes our calculation for new dollars added. That was part of the language I wouldn't have thought needed to be touched in the amendment. But looking at the doc, yes, that actually is something that needs to be explicitly stated and addressed. And so I knew enough to get it the ballpark, and then the tools helped me hit the home run. Now, again, it hasn't passed yet, so it's not a home run. Um, enough to make it playable.

Ben Webb 32:10

Yeah. And I think kind of going off of a segue kind of or a little bit of a tangent there, I've seen a couple of lesson plans this year that are AI generated. So part of what I always kind of go back to with my my kids that are an oral com or a lesson plan that is AI generated, okay, cool. You've made this outline that is made by someone else. If you don't take ownership over it, you don't give it life. If you don't put any soul in it, it's not gonna go well. Yeah. So like the prompt can only go so far. Now we get to go in to Springfield in May and really try to sell it, give it life, give some soul. Tell the story. Yeah, and I think is I think that's the important piece is this is legislation that impacts real humans. Um, same way that when we're doing graphic design for like a poster for a show. Can the AI make something faster than me? Sure. But I I get such you know gain from doing that artwork, working with the kids on it, and and telling a story through that.

SPEAKER_01 33:08

Yeah.

Ben Webb 33:09

So it's it is also about the journey and the process, not just what is it output after 17 seconds of processing what I've told it.

SPEAKER_01 33:17

Yeah.

Ben Webb 33:19

You get to ask the last question. Here's one more I want to ask you though. Okay. What guardrails do we need in schools for AI? I I I appreciate that you brought up the concept that we have to test it, figure it out, and then implement it. So, like the FDA approved piece. Um, some places are doing like a five-year moratorium or suggesting a five-year moratorium on use of AI in schools because we haven't figured it out yet. So, like, are there certain guardrails or are there certain um components there? I always emphasize the academic integrity piece. Like, I want to, I want kids to know this device, this thing could cheat for you, but I want to show you why it's so significant for you to place some steps along the way where you're making sure you show me, no, this is all me. Yeah. So what do you think on that with like guardrails and screens new? I think the best and worst thing to ever happen with a Google Doc is version history. So I can always see what the kids did and where they might have made the mistake.

Andy Goveia 34:23

And when they copy and pasted two old paragraphs from somewhere else?

Ben Webb 34:26

Yep. Um sometimes it's just from Grammarly.

Andy Goveia 34:31

I well, and that that you know sometimes we do love grammarly. Um that's a great tool. Everyone should have grammarly. Speaking of just give everyone grammarly, always every YouTube video I ever pull up. Oh, just give everybody grammarly. Um, I'm gonna this is a cop-out answer. Okay. Slightly, but I think it's it's meaningful. I think guardrail-wise, you have to have cons a good one of the best guardrails is as much consistency as possible for students across teachers. And I there are, and again, there are some classes, like in a makerspace classroom, going crazy with it and trying out all of the different things, a hundred percent. Right? Like if you're in an AI class and you're studying AI, running it through the ringer is what you should do, right? In a content area class, like we've talked about today, no, you can't just use it unfettered. But if that attitude of responsible use, ethical use, if that is consistent, if the so maybe what I'm trying to say is if schools are gonna implement it, having a slight as much as you can of a school-wide culture of you get to decide what it looks like in your classroom, but we don't compromise on this, this, and this. We all agree we're gonna use it ethically and cite it as a source if it's used. We agree it is going to help us, not replace us. Like, and we haven't done this necessarily. So I'm not speaking like, come see what we've done. I just think where it works best is when kids see that there's a value system to how it's used. I think it's where the human piece comes in. So the best, here we go. To answer the question, the best guardrail is a a set, I think a set of values that your building, your department, whatever, can all agree on. Not out, not application, not use, but hey, our philosophy is going to be as consistent as possible in these two or three areas.

Ben Webb 36:34

And and that's where I would say at UhI, we did start the year saying this was our uh PLC from last year. This is the philosophy that we saw from this university, which we would like to say that is what we agree to. That's what we think. And we said it wasn't AI generated, it was from this university, this is our AI philosophy. And we brought it to the whole school and said, we think this is significant and this is what we should have going forward. And then it went back to the teacher to then say, okay, what am I gonna do to adhere to these kind of mission critical values? And in some cases, it's I'm not gonna use AI in the classroom at all, but I'm still gonna adhere to these values about how it can be used elsewhere, how it needs to be ethically um cited if it's being used, how we're acknowledging the implications of how it must be a part of the lives of our students going forward and for the teachers. And we've we kind of had that conversation to start the year. Yeah. Um I think we also kind of started there after we went to the beginning of year, like kind of meeting. We had like a town hall where the whole faculty was together, and

Guardrails Through Shared School Values

Ben Webb 37:45

we were kind of told that teachers who ignore AI are gonna be replaced by teachers who know how to use AI. And I think that's that's another hot take that is it's being discussed. I think I'm gonna push back on that because the teacher who doesn't use AI, if they have reasons, if they're justifying their reasons, no beef. Yeah.

Andy Goveia 38:12

I would modify that statement.

Ben Webb 38:13

Yeah.

Andy Goveia 38:14

To say the teachers who are gonna get hired will be acknowledging AI, and the teachers that don't acknowledge it, like if you're looking for a job, the edge in getting hired. I don't because I and I think maybe to I maybe that's what our speaker meant. Like if you're looking for a job you're looking to hire someone, I think you're gonna try to hire the person who has a deeper skill set, and that would include AI at this time. It's like 15 years ago, do you have did you have a Google experience? That was the big like, did you know how to use Google Docs? And if you did, you were like, oh, right? Advanced. Five, six years ago, it was, are you familiar with one-to-one? Right? And now, or Canva or whatever the you know what I mean? And now it's what's your AI background? What do you know about it? Like, so I th I'm gonna I'm gonna give the most a very generous interpretation there that thinking about new hires, you're gonna go with the deepest skill set. And if AI is in that skill set, it's gonna make you a more attractive candidate. Nice. I think.

Ben Webb 39:09

Final word goes to you. Final word, final question. Up to you.

Andy Goveia 39:13

Here you go. Ready?

Ben Webb 39:14

Ready.

Andy Goveia 39:15

What worries you about AI adoption right now in schools?

Ben Webb 39:22

The worry that I have with AI is as someone who tested iCloud collaborators, so I I had like 24 different users on a beta version of iCloud Docs before we had Google Docs, and I basically broke the server.

Andy Goveia 39:41

Um Congratulations.

Ben Webb 39:43

I am I was very proud of it.

Andy Goveia 39:44

Probably why Tim Cook is stepping down.

Ben Webb 39:45

I I know. I'm very curious to see where things go. Um so, like, as much as I want to be advanced in testing new things and trying them in the beta, I think we are treating too much of what's happening in AI right now, not as the beta, which is going to have bugs. We are assuming that it's always gonna work, it's always gonna be the best. And I think too many people assume it's good enough, it's right, it's gonna always be right, and they're losing that discernment for whether or not it's supposed to work. And I think too many people are ignoring the very bottom of the screen where it says this AI can make mistakes. I don't the same way that there's disclaimers at the bottom of advertisements, too many people are forgetting that little piece at the bottom. And I think we are as educators uniquely positioned to remind kids, to remind colleagues, to remind future teachers that it can make a lot of errors. And so you have to become better at discerning what are those mistakes, what are those, what are those problems, and then go from there. You have to be in the driver's seat if I can go back to that metaphor. Yeah. Is that a good enough answer?

Andy Goveia 41:01

That's I that's we're we're both so optimistic about it. And we're users of it that it's I think it's interesting just to acknowledge, like, hey, you can be an enthusiastic user of it and still go, ooh.

Ben Webb 41:12

And and I think that's the big thing is depending on the day, I am either, yeah, Terminator's coming, or oh no, this is gonna be phenomenal.

Andy Goveia 41:19

Yeah.

Ben Webb 41:20

Cool. A few more episodes before we're out for the whole year.

Andy Goveia 41:23

Yeah. I think a couple quick things, maybe just we're hitting, we're recording that. I know people don't like to know cut timeline. We're recording this at the end of April. Um, and so it's probably coming out pretty quick, actually. So this one might be really timely. Um, I think it's really important to

What Worries Us And Listener Requests

Andy Goveia 41:40

shout out what you're gonna maybe hear from us here going into the end of this year. So you're gonna get um an episode or two with some chats with some of our colleagues talking about their year, how it's gone for them. Um, if you stick with us this season, you're gonna get an episode with some of our new staff around the buildings this year, hearing about what their take was on coming into our system. Um, if we can get it finally figured out and coordinated, you're gonna hear from some of our graduating students about their time here and what it's like to transition onto what comes next. So you'll hopefully hear from some eighth graders coming to high school and then some high schoolers on their way out into the world.

Ben Webb 42:17

Yep, into college.

Andy Goveia 42:18

Um, and then we're hopeful to do one closing episode with you all as well. Um, but here's what we need from you all. If there's things you want to hear, and I put this on Facebook and Instagram and no one responded. Our 120 followers, come on. Um, we need to know what you want to hear. Um, doing this has been just a joy for us to do. So much so we wish we had like a day a week to just do this. Um, it's been a very life-giving thing, I think, for the two of us to get to do these chats, especially with people. Yeah, these conversations with ourselves are also phenomenal, but it's just there's, I mean, the energy of a conversation, and I think you've heard it on some of our episodes this year. Um, just the energy you get from hearing from what the the great work that people are doing in our system and across the university. And so if you have something or someone you think we need to talk to, you want us to feature, we have a clean slate next year. We have 36 weeks of a school year. We'd love to put out 20 or so episodes next year. We need to know what people want to hear and want to listen to. And so reach out, message us on the Facebook, the Instagram, shoot us an email. An email, shoot us some emails. Um, we want to hear what you want to hear so we can get the people in this closet we record in around the microphone to share the great things going on across our two schools.

Ben Webb 43:50

I love it. All right, folks, have a good week. We're almost there, almost summer.

Andy Goveia 43:54

Woo! Live in the Lab Schools is a production of the ISU Laboratory Schools, recorded on the campus of Illinois State University. Connect with us on Facebook or Instagram at Live in the Lab Schools. We'll see you next time.