NEW Applied AI for Educators

Good afternoon everyone. Well, I know it's afternoon here. I don't know where everyone else is located. All right. While you guys are all getting logged in and ready, I'm just going to go ahead and introduce myself. I am Danielle Santilli. I'm a PD specialist here at CodeHS and I work with educators on building practical AI skills that they can use in their classrooms and workflows. So today's session is 90 minutes and we're probably not gonna get to everything AI in this 90 minutes, but we're gonna get through as much as we can. And hopefully by the time we wrap up, you're going to leave with a real tool you built yourself in a framework that you guys can start applying right away.

All right. All right. So, I'm going to go ahead and launch a poll just to kind of see where we are all at as a group here. There's about just over 20 of us in this space. Go ahead and launch. So, just curious, you know, I know we're all in our different AI journey. So, where are you kind of currently with using AI in your work? So, we'll just take a moment or two here. Go ahead and share where we are at. All right. And if you want to, you are welcome. The chat is available for everyone to use. So if you wanted to just kind of say hello in the chat as well, share where you're joining us from. I always like to see where everyone is. I know we have a lot of a lot of people who are part of our learning community here.

All right. Wow. All right. Okay, so we have some Memphis, Dubai, Iowa, California, Boston, hot Arizona desert. Oh, I feel like it's hot pretty much everywhere right now. It's hot. I'm based out of Detroit, Michigan, and it is. We're on a heat wave right now. They just sent out the heat advisory to everybody. All right, Maryland, Florida, Texas, Los Angeles. All right. Awesome. Well, thank you for taking time out of your probably summer vacation, right? Everyone's kind of enjoying some time off from the school year, the hectic school year. Summer. Yeah. Yes, definitely. And hopefully you have something fun planned after this to enjoy your summer evening, staying out of the heat too much. If you are out in the heat, maybe you're at a nice pool or swimming in a lake or ocean depending on where you're at. It's how I usually like to spend some summertime. Ooh, Toronto, Canada. All right. Yeah, I've been to Canada many times being based out of Detroit.

All right. Well, thank you so much for sharing. Let's take a look at our looks like most of us participated in our poll. So, I'm going to go ahead and end that and share out the results. Okay. So, it looks like about half of us are using it regularly. It's already a part of your workflow. That's awesome. I think that's great. You've already kind of embraced where we're going where, you know, the tools that we have at our disposal. Some people about 25% tried a few times, not super consistently. Some people curious, some people a little bit skeptical. That's fair. I think, you know, we should be skeptical when there's new tools out. Make sure we kind of get knowledgeable about them before we kind of implement them in our workflows or, you know, with our students. Oh, Egypt. That's awesome. Wow. Dubai, Egypt, United States, Canada. We're we're international here. All right. Well, thank you so much for taking some time to share where you're at and share both geographically and also where you're at in your AI journey. All right. Going to go ahead and go back. All right. So, let's go ahead and get into it.

Right. So, want to start with something that might push back a little on how AI professional development usually goes. Most of the time when educators hear about AI, the message is use it more, use it faster, integrate it everywhere, right? We just want to, you know, there's new tools, we need to start using them, you know, getting getting them in our workflows in our classrooms. But, you know, I want to offer kind of a different frame. The teachers I've seen get the most out of AI are not the ones using it for everything. They're the ones that kind of know when to use it, when to kind of push back or when to just not use it at all. And that judgment, that discernment is what we are building on today. It's not just, hey, a list of tools. We're not trying the newest AI gadgets just to use them. This is, you know, not a tutorial on how to use a vast array of different AI tools. It's a way of thinking and it starts with one idea that I want to kind of hold on to throughout this session. AI adoption is not just a technology decision. AI it's an instructional decision, an ethical decision, and a professional one. And everything that we do and we touch on today is going to kind of stem from that idea.

All right. So before we can talk about how to use AI responsibly, it helps to have a shared vocabulary for what AI actually is because not all AI behaves the same way and that matters a lot for how you evaluate it. And what's interesting is I just got back from doing a PD in South Carolina and when I talked about there being kind of different kinds of AI, like everyone was like, "What are you talking about?" I mean, like AI is ChatGPT, right? Well, AI is more than that. AI is kind of an umbrella term for a lot of different kinds of tools and technologies here. So, we're going to kind of break it down into these kind of four behavior types. All right. So, most of the AI tools educators encounter fall into one of these categories. Assistive AI supports a task you're already doing like grammar check. Yes, grammar check is AI. That's something that's been around for a very long time. And AI has been around for much longer than just, you know, 2022. And then auto scheduling, helping you organize your notes. Generative AI creates new content. So, that's the big one. That's like the ChatGPT that generates text, images, audio, or lesson materials based on prompts. And these are really what we think of first. When we hear AI, we think of we just naturally kind of go to that generative AI. Then we also have predictive AI that uses data to forecast outcomes. So you might see this in platforms that flag at risk students or recommend next steps. Then we also have adaptive AI which adjusts based on how someone interacts with it over time. This could be like a personalized tutoring platform or I often think of those testing that we'll do with students where if they get it right the next question is harder than the last and it just kind of adapts to where the students are at.

All right. So one thing also to keep in mind is that these categories are not always perfectly separate. So a tool can fit into more than one type depending on how it is used. So for example, a teacher might use an AI tool to help draft a parent email or use it to, you know, use that in a communication capacity. So in that case it's assistive because it's supporting in communication. That's something the teacher was already doing, but it's also generative because it's generating new text. So that overlap is important because we don't just evaluate a tool by its name. We evaluate that tool in that specific moment. So let's hold on to these four types because that's what we're going to be reflecting on in the next section. We're going to come back as that foundation of the evaluation framework. So this is a two-minute level set, not a full lesson. All right. So just kind of jumping in to those main pieces. If I am confusing or if you need any clarity please, you know, if you could throw out a question in the chat, definitely want to create a collaborative session for you guys.

All right. All right. So here is the five question framework that I kind of developed to help kind of analyze tools because we've all been in that scenario where, you know, maybe a colleague says, "Hey, you should try this tool, it's amazing," or maybe it's even coming from, you know, above, "Hey, we're going to be implementing this new technology piece and we need you to try it out." Or sometimes maybe even a vendor emails you with a free trial offer like, "Hey, try this new tool and see what you think." And, you know, it's natural to be curious or maybe even excited like that sounds like a great thing, but we should know that before we use any kind of tools, whether on our own or in front of students, especially if we're going to enter student information or any sort of educational context to it, we need to maybe be able to kind of answer these five questions. And that's kind of what this framework is for.

Which one of these is most appropriate in teaching, Gemini, ChatGPT, or Claude? Okay, so that is a great question. That question is probably one that your administration probably has some direction on. So, I've worked a lot with different school districts and many of them will tell me this is what's allowed and this is what's not allowed. I have noticed that Gemini tends to be the one that's a little bit more school friendly, mainly because there's a lot of school districts that live within that Google verse and Gemini is so deeply embedded now in those tools that a lot of schools are like, "You know what? That's going to be the one that we allow." However, you know, it does depend on your school. Magic School is a very popular AI tool. It works a little bit differently than a typical chatbot. And I do believe that there is that's kind of like a premium with Magic School where there are certain things you can do for free and then other aspects that you need to have that pro account for. So I definitely feel like it does depend from school to school, but most of the time I'm actually hearing it's either going to be Gemini or Copilot. It just depends on what universe your district works in. But excellent question. And I've used all three of these and find that, you know, some work better for different kinds of tasks. I like Gemini, especially when it comes to creating images and stuff like that. I find like Gemini is on point with that. However, ChatGPT has also made big improvements in their image generation, too. So, it really just depends on what tools you use and your own preference. So, yeah.

Okay. So, AI, the AI stuff within CodeHS, we have a lot of AI tools that are just in the platform, so they're not directing people to third parties. So, we've kind of built our own tools within the platform and made a lot of updates within that. So, it's not necessarily like we have built-in, you know, ChatGPT. It's kind of like our own thing. Yeah. Gemini, I feel like Gemini is a popular one, especially since it's embedded so well in the platform here. But before we dive too deep into which chatbot is better, let's talk a little bit transition back to this framework here and kind of talk about, you know, maybe we can even think about some of these tools in terms of this framework. All right. So the first question is what type of AI behavior is this? So I just kind of talked about those four different categories and if we look at those, you know, is it assistive, is it generative, is it predictive or adaptive, or, you know, which one or is it a combination of two of these, especially with assistive and generative kind of being so closely related and overlapped. And then we have our second question here. What inputs does it use? And this one's important too because we have to be very careful about what kind of information this AI in this tool that uses AI is asking for and not necessarily either like I'm giving it text and it's using that text but also like are you having to create logins for yourself or your students? What is it looking at? Is it collecting sensitive student data or even your own data that you don't necessarily feel comfortable sharing?

Another thing too is, you know, a lot of these AI agents might use information to train its AI. I noticed that a lot with free tools that they don't have that little message at the bottom that says this data can't be used to train models. So, I would definitely see if your school district has an AI tool available that they're paying for cuz then most likely it won't be used to do that. I know like for example, if I'm using my work ChatGPT, it does have a little thing that says this data is not used to train our models, but when I use my free version that I don't pay for, it doesn't have that. So that is something to kind of be aware of, is my information being used to train the model? Another question, question three, is what output does it produce? All right. So what is it that it's giving me that would be very helpful? So is it drafting sub plans or is it creating images or is it giving me a great insight into where my students are at in their content understanding? So it can give you, you know, sometimes we decide like the outputs are significant enough that maybe some of the inputs we're willing to put in.

And then question four, most importantly, what risks exist? So, we're talking about, you know, hallucinations, biases, privacy issues, and then of course when we're talking about using AI with students in particular, that overreliance piece. Oh, I like that. Thank you for sharing. Okay, so in the chat, asking questions to different kinds of AI and asking for critique and then going back and forth, right? I've done that too. You know I will use Claude to generate something and then put it into chat and say, "What do you think about this?" or "Can you tweak the language a little bit?" So yeah, kind of experimenting with different tools is helpful. So that's an excellent thing to share.

All right. And then we have our last option here which is what level of oversight is required? Number one takeaway from if you remember one thing from this session it is that human oversight is required when you're using AI. It's more about what level of oversight do you need, but human oversight is always necessary. You absolutely need to look over even if it is just that, you know what, it's just a quick review. Maybe you're just using it to brainstorm some topic ideas for a lesson hook or something really easy like that. You know, okay, I'm just going to read over what they said. Okay, I like this one. And then you move on, right? Just a quick review of what it said and then move ahead. But some things might involve more careful fact-checking. If you're looking for information that you're going to be teaching with your kids, you want to make sure that it's not hallucinating events or dates or studies or anything like that, right? And then also last level is it just is it not appropriate to use because that risk level is just too high? So kind of understanding those questions, and it's not when I say what level of oversight is required, it's really not a yes or no question. Yes, it needs or a no. It's because the answer then would always be yes, oversight is required. However, it's that spectrum. Exactly how much oversight do we need to be looking at?

All right. So, yeah. So, here's kind of our spectrum here, the risk spectrum. So, now that we kind of have our five questions, we want to turn it into a decision. We understand that all AI use carries some risk, but once we kind of can see the clarity, we can lose some of the anxiety about it. So maybe if you're doing something on the low end like brainstorming or drafting or maybe just formatting a paper making sure that it's in the right format or generating a slides presentation. You're just giving it the content. You just want it to generate and organize the slides for you. You know, probably more of a low risk. You would just kind of want to review it, make sure it looks like it passes the human test and then go on from there. Or is it a little bit more medium risk? And I feel like a lot of what we do kind of involves that area, too. You know, if you're summarizing content, generating those explanations, you're creating lesson materials, especially, you need to make sure that you're really looking through it, having that nice review, making sure it's factually accurate, that it's not missing some nuances, especially when we go to summarizing content.

And I know a popular use for teachers is taking a text and maybe changing the text to be more reading level appropriate for different levels of students. So I've heard that's kind of a way that some teachers will use AI. And I think that's a great use, especially if we're trying to differentiate for different levels of learners all in one space. You have 30 plus students in your classroom all at different reading levels and we need to quickly create materials for them so that they're able to be successful in our classrooms. But we just want to make sure when we're using it to create summaries or to rewrite text in a more reading level more appropriate to them that we're not losing the facts, we're not losing the context, right? That it's all still there. So, we are making sure we're reading through it and that it does still check off all the boxes. We don't lose anything in translation, so to speak. So, yeah, just but it's definitely very helpful, right? Differentiation. AI can be a huge help with that. But just kind of making sure that you're still reading through it. You're making sure that it's still factually accurate. It doesn't have any kind of biases. I also noticed that with image generation, making sure that you're not playing into any kind of biases or stereotypes that exist, right? We want to make sure everything is kosher before we send it out. All right?

And then of course we have the high-risk. So, high-risk student evaluation, family communication, official documentation, you need to have that really high lens if you're going to use AI at all. And when we're talking about these things, and I'm not saying you should never use any AI tools if you're looking at student progress. I feel like I mean, I put student data in spreadsheets when I was teaching. I was a teacher for about 14 years before I started working for CodeHS. And I always had to give my student growth for part of my evaluation, right, to prove I was doing my job. And yeah, I would try to make spreadsheets and show whatever and charts and all that. And with AI, you can do it easily. And I'm not saying you can't still use AI for that, but maybe you're just using it for a portion of that. Maybe you're anonymizing student data. You're not including any names or anything. You're just showing the trends and then using that piece and then putting it together in the bigger documents. So, there are ways that maybe even with these high-risk documentations or things that we're doing, we're just using it for one component of it. So it's kind of that balance of what part is AI a help and where would it be a firm no, that's not appropriate use. And obviously that firm no is when we're talking about student personal identifying information. We want to make sure that's never something that we share with AI unless of course there is a specific AI tool that your school is telling you is HIPAA compliant or FERPA or whatever. I always get the medical one and the school one confused. But, you know, I know that they do have some programs out there that I think are becoming a little bit more secure, but when it comes to general AI, no, we don't want to share any of that information with that. So, you know, AI supports your personal judgment. It does not replace it. Human oversight is always always necessary. All right. So those are kind of your three levels of risk analysis here.

All right. So what we're going to do, yay, we have another poll. So I'm going to go ahead and launch another poll. It's just going to have three questions here. And you can kind of decide, you know, what would you trust AI to do? Would you trust AI to do this? So go ahead and launch that out. So just three little questions here. Would you trust AI to write a first draft of a lesson plan, summarize a student's progress for a parent email, or recommend whether a student needs intervention support? So just kind of see what you think.

All right, looks like most of us have kind of responded here. All right, so let's go ahead and I'm going to end this poll and let's take a look at what we all thought here. All right, so for the first one, yeah, I think that a first draft of a lesson plan looks like most of us kind of were like, "Yeah, with edits." Yeah. And another thing too, I want to note that it does say first draft, right? Would we trust AI to make a first draft? You know, maybe we feed it some ideas and we want it to kind of structure it into a lesson plan. Maybe, yeah, if you gave it enough context, maybe a quick review is all it really needs for the draft before you kind of go in and finalize it or whatever. Or yes, with edits. Yeah, kind of tweaking it and making it substantial so you're able to make it into more of a final product. Still a couple people say no. That's fair. Next one. Summarize a student's progress for a parent email. Looks like a lot of us say yes with edits. Okay. We're looking at I would say with that one, yeah, if you're looking at that progress and you're not including any of that personal identifiable information into the AI, you know, and just kind of making it nice vocabulary for parents, very professional, right? Or sometimes, you know, if you're very frustrated with a student, maybe it could soften it down, right, to a nice a nice tone there. All right. And then our last one. So recommend whether a student needs intervention support. Okay. So here we're a little bit more mixed, but the highest percentage here just just south of 60% said no. You know, when it comes to interventions, maybe we can use it as a piece to help us see trends and see where that student's at. But I think it's always up to us as the adults to kind of make that call. Does a student need recommendation? It needs some intervention. Maybe we look at data sets that we got from AI and from different places and that might influence our decision, but we should be making that kind of decision on our own for sure. All right. Awesome. So, I'll stop that one here. All right.

So, okay. So, we kind of talked about your instincts. You guys all kind of just based on what we talked about in your own experience were kind of able to decide how you feel about things and now you have a nice little framework that you can show to back that up. So, it kind of goes from that feeling of "I'm not sure about this," that gut feeling, and it kind of breaks it down. Oh, it just said recommend not make a decision. Absolutely, that is true. Yeah, it could make that recommendation, point you in that direction. Yes. All right. But what's nice is you take that gut feeling of how you feel about it and then you apply those five questions and kind of analyze it and that can kind of help you make that informed decision or at least you're now better able to kind of articulate why you feel that tool is or is not appropriate.

So, what we are going to do next is I created this fun little fake AI tool that I'm going to throw the link in the chat here. And so we are going to practice applying this well, I should say you guys are all going to practice applying this five question framework to it. And what I want the main takeaway is, you know, this actually sounds like a tool that could exist and it's going to prove that there is not necessarily an easy right or wrong. So, let's take some time. I threw the link in the chat here. Please let me know if you are not able to access this. Once you go to that site, oh, it's not working for you. Let me try this to let me see if it works for you. Sometimes if I change the ending. Okay, I'm going to throw this link in the chat. So, if you weren't able to get the first link to work, maybe this one will work for you. Okay. So, try either one of those links and see if you could get one of them to work. All right. Oh, good. The new one's working. All right. Yeah. Sometimes at codehs.me codehs.xyz, same thing. All right. Oh, no. All right. Okay, good. I see some thumbs up. All right. Well, if you are able to access it, kind of go through just click this button. It says start evaluating now. Just okay, good. Just going to go ahead and read that intro and then you'll go through the five questions here. So, this is called Scribe Mind, an AI writing coach for K through 12 classrooms. It says, "Scribe Mind gives students instant personalized feedback on their writing, grammar, clarity, structure, and argumentation. Students pace their work and receive line-by-line coaching. Teachers access a dashboard showing each student's writing progress over time. Scribe Mind learns from how students respond to feedback and adapts its coaching style accordingly." All right. So, it says unlimited student feedback, teacher progress dashboard, works on any device, Google Classroom integration. Then, of course, a little caveat here. There's 30 students free. After that, you have a charge. Sure, we're all familiar with seeing stuff like that. So, I'm going to give you guys a couple minutes to go through this and then we're going to go ahead and discuss.

Okay. So, if you're having a hard time navigating here, so essentially the first question is what AI behavior type is it? So, you're going to kind of look at the choices and you're going to click on the answer that you think best supports. So based on this description of this pretend AI tool, do you think it's assistive, generative, adaptive, or both assistive and adaptive? When you click on it, it should direct you to something else. So for this example, you know, what choice? Oh, okay. Yeah, no problem. No, I want to make sure everybody understands how it works. And then if you do finish this, you could just maybe throw a thumbs up in the chat so we kind of know where we're all at. I want to make sure everyone has enough time to kind of go through this so we can have a thoughtful discussion after.

All right, looks like a number of us have kind of wrapped through it. If you're still working, that's totally fine. Just go ahead and work through here. What did you think about this tool? Is there anyone who went through and felt very strongly one way or another? Or do you feel like this is kind of a realistic scenario here? Or what are our thoughts? Yeah. Yeah. I thought so too. It really is. I feel like this is definitely something that probably already does exist. This is probably just something that's already on the market. Then going through that framework, did we feel yeah, it made you think about other tools that use similar kind of looking at things in a different lens. I think kind of helps us better understand what we're using and how our students might be using these tools. Yeah. And what I like about this is yeah, it does seem like a great tool. Yeah. The English user is the biggest anti-AI, right? For sure. What I liked about this is yes, I feel like it creates a very realistic tool scenario that I think we can all agree that we have seen things like this. I think it helps to kind of look at I like that.

Okay, so the first one here, you know, it does show that some of the takeaways on how that could contribute to risk as you go through each question and how that kind of escalates because of the I think the escalation here for the inputs is really because it's tracking student progress. So that is student data that it's collecting. I think that's kind of why it gave you that that red alert here. But I think it can help both students and I really like that teacher dashboard could be really helpful. And then obviously all of that could really realistically I think that overreliance piece, especially when it comes to writing, can be something that we struggle with. And then I think that level of oversight, one thing I really liked here is this risk level thing. So, yeah, there are like as we were kind of going through that framework, we were seeing, you know, some things were those red those red risk higher risk factors because it is collecting student writing samples, right? Because we know it's using that information to provide intervention and feedback to that student. So that is data that's being collected. But you could look at the other side. It could be helping them with their writing. It could be helping differentiate for student needs. Maybe some students really need that additional scaffolding and support. And it's hard for us, you know, being the one teacher in a classroom of 30 plus students sometimes, that it's hard to provide that scaffolding and intervention. And if we have a helpful tool that can work with that, you know, we just kind of have to look at things and mitigate, balance the risk-reward.

And I feel like having a framework might help us evaluate those tools and kind of see all of the risks and then see the rewards and kind of decide where we're at. And then I also like down here it kind of says, you know, this is something that you probably need to get admin say on it. This is probably not a tool that you'd be able to just start implementing in your classroom. And then kind of I think that this is helpful, too. If your admin is skeptical, walking them through this framework. I do think that there is a difference between saying, you know, going to administration saying, "Hey, I have this AI tool. I really like it. I want to use it." And then they're like, "Uh, you know, what do you mean? I need to do more research on it," and then they're not going to, right? They're busy. Not like we aren't, but if you're like, "Hey, I have this tool." Yes. Yep. We are recording this definitely. So, yes, this will be fully recorded. Oh, and I'm glad that you are enjoying the training thus far. Thank you. Yeah, but I do think that, you know, having a framework that you can go to an administrator and say, "Hey, this is a new tool that I was looking at and doing my evaluation of the tool, here's what I'm thinking," and you know, just kind of I think when you present it in a way that's like, "Hey, I went through this framework, I really analyzed it, I looked at the inputs and the outputs," and then obviously if you're taking it to admin you believe that the risk-reward is worth it, you know, just kind of I think that can help get that conversation going in a positive way. I know in my experience, a lot of times there's AI tools that come out and then people start using them and then all of a sudden they're like, "Hey, by the way, you're not allowed to use this tool anymore. It's blocked." So maybe being proactive and saying, "Hey, there's this tool. I ran through this framework. Is it something that we can use?" might help that conversation.

All right. Okay. I'm just wondering if this would be helpful to students. Create additional dilemmas. Yeah, I think that yeah, creating multiple different dilemmas and frameworks that people can kind of practice going through could be helpful. Yeah. All right. Any other kind of questions or comments about the framework before we kind of move on to our next phase of our time together. Jump back to the slides here. Okay. Any use for computer science? I mean, there are tons of AI programs that are specifically designed for computer science or that you can use for computer science. I think this framework would work for any sort of AI tool available. It'll yeah. All right. So you teach AP computer science, generative AI is not allowed. Well, with that, I mean, maybe you're not allowing students to use generative AI, but maybe there are tools that you use that are AI. Maybe you're using different tools to create different classroom materials that you're using with your students. And you need to make sure that it is stuff like that. I think so. I think that AI can be a really great tool in computer science. I think that those are all assets for helping learning, especially when students are really struggling. Maybe using AI to generate hints or AI to help point students in the right direction to find bugs in their code. Or maybe even understanding code that may be more complex to kind of explain it in simpler language. I can see how AI would be very helpful, especially if you had a platform that allowed that your school would allow students to use.

I see such benefit as long as we're teaching students to use AI appropriately. I don't think that there's anything wrong with allowing them to use it in certain ways for certain activities. So it is tough when they do just put that firm ban on generative AI not being allowed with students because yeah, it definitely can be very helpful. I know even, you know, we at CodeHS, those of you who use our CodeHS curriculum, if you have a pro account you'll see that we do have some different AI tools that we allow teachers to use, teacher-facing tools. That it is that balance. Are our schools going to allow their students to use any AI features? But yeah, I personally I feel like that kind of leads actually that leads directly into kind of what we're going to be talking about next, which is using AI with students, right? Wait. Oh, yeah. I was right.

Okay. So, when we are using an AI tool, if we're deciding it's worth using for ourselves, to develop materials, or using for our students, I think a lot of it is just kind of giving our students direction on appropriate AI use. We know that students are using AI, period, end statement here. Students are using AI, whether we block it on our school devices or whatnot. You know, it's like people have been cheating since the beginning of school, right? People are always finding ways to get around doing the actual work. So, people are going to use AI whether we like it or not. So, I think the main thing is having there be like an AI use policy. And a lot of districts will have some blanket statements regarding what students are and are not allowed to use as far as maybe tools go. However, I feel like it comes down to the teacher to kind of decide what activities and what they think is acceptable use of AI for their classroom, for their content area. I think AI use for computer science will look different than AI use in an English class or AI use in math. You know, I feel like we use these tools in different ways for our students to help them learn or to help them understand the content. And so I think that we should all kind of be transparent with how we would want our students to use it. And I found that this kind of framework might be helpful.

Yeah, people have been cheating. Yes, AI only lets students Yeah, I totally agree. Yeah, I mean AI is just making it maybe a little bit easier and faster, but we we're finding ways around it. And I think part of it is kind of deciding, you know, what are activities that are blanket like no AI? Like, "No, I need to see what you learn and know how to do on this activity. The student's thinking is required, they need to go through that productive struggle, or they need to showcase their understanding of a topic. No AI." But maybe we do allow some activities or even parts of bigger projects to have that limited AI. You know, maybe we're going to say, okay, so in this project, for this step, in the brainstorming step, you are able to use AI to help you brainstorm. Or maybe after they finish their first draft, they can use AI to help them do some of the feedback on checking your work or, however you decide to see it, but kind of being very transparent. All right. So, here is how you can use AI to help you either get started or use AI, right? I love that. Show and explain your work is no longer just for math class. Absolutely. I feel like that's just part of it now.

And another piece too that I think is helpful is if you are having that stage of limited AI or AI-supported, where AI is part of how you're doing that activity and you're not grading what AI produced, you're grading the choices you made with it. I really like that. It's like, "Yes, I'm wanting you to use AI with this. I want you to use AI for this project or this activity because what you're doing is you're now going to analyze the results or you're going to share why you used it or why you think it..." I don't know, you can kind of bake it into the assignment where now they have permission, but they're they're using AI in a way that's acceptable for the assignment or for the learning or they're learning how to use a tool the right way, right? We need to teach students how to use AI appropriately. You know, when we're using AI, we should be being transparent. I used AI to help me build this tool and this is why I did or this is how I did it. You know, I was always very transparent with what I was doing with the students and why.

When I started teaching computer science, I was completely new to computer science. I don't have a degree in computer science. I was a music teacher. I was an elementary music teacher who then moved up to teach high school computer science. I was very transparent with my students on what I was doing and how I was learning. And that's just kind of how I was able to get through it. We're all learning together. We're all learning this new tool together. We're all learning how to use AI. This is how I'm using it to be successful. Another thing that's important too is I feel like, you know, as educators, we know that AI is becoming a part of life after school, right? You know, these students are going to enter the workforce where AI is becoming part of it. And if we just shelter them and they're only using it on their own in however they think is appropriate, which is probably just typing the directions to the assignment and having it create a result for them, we're not really giving them the tools and the how-to on how to use things appropriately. So I think, you know, having it be very clear and transparent on what assignments what is it that you're I think thinking about what is it that you're truly wanting to assess from the student, right? What is important? Is it something where it's, "I need to see their thinking," well then no AI and they explain out their thinking. Or maybe it's, "I want to see how they understand the concept of the code," so I'm going to have them use AI to create this code, but then they have to analyze it. They have to say in their own words what's going on here. And maybe it is an unplugged assignment. Maybe you generate the code and then they're working with partners and they have to write out like the explanation somewhere where it is more where you're actually kind of assessing their knowledge on that topic. So, I do think that it is helpful to build in different activities that some do but some do not require the use of AI or at least allows them the option to use it. Maybe I shouldn't say require. However, if you have them analyze code that AI generated, that would be required. But I do think that that is that is very helpful.

And then I think that, however, with a caveat, I think that yes, we should be having students use AI appropriately for different activities in our coursework. However, before a student uses any AI tool, I think they really need to have an understanding of these three things. And I think that some people might think that they have an understanding or kids might be like, "Oh yeah, I know hallucinations where it makes up facts." And that's true. AI definitely can create false information. I think talking about it as a real thing, having them understand that, yeah, AI is not always going to tell you the truth because AI is a prediction machine. AI doesn't know anything. AI does not understand. It is just looking at vast amounts of content and playing a probability game of what do I think probabilistically speaking—that's not a word—but yeah, what is it really saying here and then it predicts its response, and so that's how it creates those hallucinations, right? Because it's not going off of fact. It doesn't know right or wrong. It's not searching the vast array of information to find the correct data. Data is just, you know, predicting the next word or predicting the next pixel in the image, right? It's very I think that's an important understanding. So it's not always producing correct information, we always have to, you know, if we're if we're using AI in a research part, I think it's very important that they have to make sure that that is real information. They should be doing research on what the AI produces. Okay, can you find two sources that validate what that AI said? You know, stuff like that where they're really looking at it, creating those activities. Yeah. Oh, good. That is important. How human brains work versus how computer quote-unquote brains work. I'm always very intentional when I use AI to make sure I'm not humanizing it in any capacity. I don't want to blur those lines, right? AI doesn't think. AI is not human. It is a tool that we use, right? So I do think that creating that clarity is important.

And the next one, all right, so the second is so important, and that's the I think it's pronounced sycophancy, right? AI is always so validating and happy and it wants you to keep talking and you're always right, you know, and it's gonna really change the way it it communicates, right? Its feedback to you is going to be based on making you feel good, right? It's going to be agreeable. It's going to, you know, it wants you to continue interacting with it. That's how it was trained. Those are some of those guidelines that the developers put in it. They want you to keep using it, right? And so that can be not great if we're talking about students using it as a tool. Students should be challenged, right? They need to have people questioning the reasoning. I did an interesting activity in a PD where I had a group of educators—I had some of them put in a prompt that had a positive lens, others had a negative lens—and we kind of compared how the output differed even though it was the exact same topic. And it is the framing of how you put things into AI will influence how you hear it back. So I think that students really need to be aware of that. Also, just from a perspective of like for their own like we don't want them to use AI because they're now they have this AI tool that's always making them feel like everything they do is fantastic. We need them to be learners and going through that productive struggle and to have that challenge.

And then of course the highest peak is overreliance. When AI does too much of the thinking, students will lose practice with the skill they are supposed to build. All right? So the goal is support, not outsourcing. So balancing that level of scaffold that we support for students or how much they're using it for different things. I will say that like, you know, when I'm typing anything in, I don't care about spelling or grammar or pronunciation cuz I know that the tool I'm using is going to fix all that for me. I haven't cared about misspelling a word in forever because I just auto-check, you know, it'll do that. I mean, is that overreliance? Maybe, probably. But, you know, so we just kind of have to decide, too, like what is the important thing that we're assessing and what's the thing that we're willing to like let go of the wayside.

All right. So, with all that in mind, I'm just going to break it down to this. When you give an activity to your students, you know, just kind of pick a tier and add a sentence why. So, I feel like if you're able to introduce and develop a framework that you think works for your content area with the levels that you think are appropriate—and I think everyone's, if I go back, I think everyone's examples might look a little bit different on what you think a limited AI might look like in your content or supported AI might look like in your content—and I think kind of when a semester starts or when the school year starts, depending on your grade level and how you see your students, you know, kind of starting off, "This is how we're going to use AI in the room. These are kind of the levels that we have." And then following through with when you have different assignments, tell them what tier it is and tell them why. You know, if you're saying, "This activity is a no AI," you can just say something simple like, "This is no AI. I need to see your thinking, right? I need to understand that you understand the concept." Or, "This is limited AI. It's part of the task, but I'm grading on your choices, right?" So, just kind of explaining like, "This is the tier and this is kind of why." And you can keep it very simple. It doesn't need to be a paragraph or you don't need to preach why you're going hard on the no AI for this activity. Just keep it simple and just be like, "Okay, this is what we're doing. This is why," and go from there. And then for obviously for bigger projects maybe it looks a little bit differently. Maybe you break it down by part. The way I think is for a lot when I was teaching everything was project-based. Teaching computer science, they were creating things. So we didn't really do any like tests or stuff like that. So everything was project-based. So I'm thinking of different projects that I did and how I might be breaking it down. So it kind of depends on the activity. Maybe it's a simple activity. You can just pick a tier sentence why. If it's a bigger project, break it down. This part you can use AI. This part, no. You know, just kind of deciding how you want to do it. Has anyone ever, you know, you can share out in the chat, has anyone kind of done something similar to this or want to share out how they've kind of created an AI policy for their students and their classroom? Cuz I remember in the poll a lot of you guys said that you were using AI pretty readily. Yeah. Because I know we learn a lot from each other. So if anyone has some things they'd want to share out before we kind of jump to our last little section of our time together.

A paper assignment to comment on a code sub. Yeah. No AI in any work submitted for evaluation remarks. Yeah, just having that firm like, "Okay, if I'm using this to collect data on how well you know a topic, I can't have you use AI on it." I feel like that's pretty that makes sense and I feel like students would do that. Students are allowed to do drafts and review prompts. Yeah. And you review the prompts. Okay. Yeah, absolutely. And I know there are some, you know, I know someone brought up Magic School. I think that there are some where you can actually see the student conversation with AI and some of those different tools. That's nice. All right. So, limited AI for brainstorming and use as a tool. Yeah. One thing I do like too is sometimes people will say like if they do use AI, part of the assignment is to explain like how they used it or how it benefited them in the assignment. Ask them to cite their sources. Yeah, I know a lot of AI tools will provide where they got that information. Just make sure that students are actually like clicking those and make sure that that is real because I've had AI give me sources or links to YouTube videos that do not exist. So, they do have to go that extra extra mile and say, "Hey, you know, AI cited this source. I went there and confirmed that information," or something like that.

Yeah. Okay. So, yeah, copying, putting in AI, and then just pasting it into the platform without even running the program. Yes. I don't think students realize how easy it is for us to kind of tell that something is AI generated. You know, there are so many ways that we're able to kind of flag it and it's like, "Come on. Like I feel like you should be working a little bit harder on this activity." Yeah, I did have when I was teaching web design, I had them I don't remember I think they were supposed to be making a web page about a topic of their choice. So, everyone's topic was supposed to be different. And then I had somebody go in and typed or pasted in what was obviously a like a ChatGPT or an AI-generated web page. And then they even kept in the bottom where it says, "I think that this should fulfill all of your project requirements." And they kept that part in the bottom. I'm like, "Um, you didn't even like proofread this enough to like, I don't know, edit the part where it's very obviously showing me that it's from AI." Yeah. But I know that, you know, a lot of I know, right? Yeah. I know that a lot of different programs and stuff, I know within CodeHS, we have a lot of different ways to monitor academic integrity and, you know, flag the pasting and all of those things, too. But, you know, even without some of these tools, sometimes it's just really obvious, right? I had a student use AI to create Python programs. They used the wrong programming language. Yeah. And they didn't even notice. They didn't even notice that it wasn't correct. How ridiculous. Generate an AI output and then they edit the output on paper with annotations to show their learning. Absolutely. Absolutely. I think that, you know, if they are showing what they're doing, they're showing those annotations, they're showing how they're understanding the output, that's acceptable AI use. That's teaching them how to use the tool the way they would probably use it in the real world, too. I mean, you know, I use AI in my job. You guys said at least what was it, 50 or 57% of you guys said that you're using AI. I mean, we're we're using it. It can be a great tool to help generate some really great material. So, we should be allowing students to use it in that capacity. Yes. Yeah. Exactly. If they if they only use AI, they can't find and explain those errors. So, I think that part of it is rethinking how we're assessing our students, how you know, what are the types of activities that we're giving our students to do. If they're super easy that AI can do them for them, then maybe we need to rethink rethink our assignments or rethink how we do them. Yes. AI and then bug and debug. Yeah, you guys are finding some really great ways to allow students to use AI in ways that'll help them and ways that they'll understand things better. If we can allow students to understand our content better by using a tool, we want them to be able to use that tool and we want them to be able to use that tool in an appropriate way. You know, any tool, whether it's AI or not, we need to teach them how to use it and how to use it appropriately. So, thank you guys so much for, you know, kind of sharing out what you do. I know you guys learn from each other very well.

All right. So, we have some time left. We're going to jump into our last thing, which is we're actually going to make something together. All right. So, we're going to I'm going to demo this using Gemini. However, you know, you can do this with other AI tools. I just know on Gemini, you can do this with the free account. I know that don't think you can create custom chatbots in ChatGPT unless you have the pro account. So that's why if you have Gemini, if you don't have Gemini, I think you can still use that as a tool and play around with it. It is completely free. I'm going to throw the link in the chat here. But we are going to be creating a custom Gem, they call it, which is really just a custom chatbot.

So, I'm going to go ahead and I'm going to jump to my other screen here and we'll kind of go through this together. And what's really nice about Gems or these custom chatbots is when you're you can get them to do a repeated task that could often be tedious or time-consuming or maybe it's something that you use AI to do already but you constantly are telling it the same directions over and over again. What custom chatbots or Gems can do for you is it can help you kind of do that. It kind of builds in the basic directions for what you're wanting it to generate and then every time you use it, it already has that information to kind of go back on or those directions to go back on. So, I've seen people use it for creating newsletters for parent communication, planning lessons, creating differentiating materials or rubrics. Maybe you already have it programmed with all of your standards. Just, you know, you can give it a lot of information and it just kind of is a great place to kind of give it direction and then it'll use those directions every time we go ahead and do it. So, we're going to go ahead and we'll build one together.

And what we are going to do is we'll be giving it—and I'll actually jump back to the slides real quick, I think I jumped a little too early—the main thing is we're going to be giving it these four things. We're giving it a purpose. So, what is it we're wanting it to do? So, I gave some examples. Writing a parent letter summarizing what we covered in class this week would be an example of one. The audience. Who is the Gem for? What grade? What subject? What content does it need to communicate clearly? The next would be guardrails. You know, what should the Gem never do? Always ask for topics before writing or keep under 200 words. Like maybe you just want it to be very concise in its verbiage. We all know that AI can be very verbose. Sometimes it goes on and on about a topic. It's like no, I need you to parse that down a little bit. So, you don't have to always, you know, do that. And of course, you're never including student names. And then the tone, right? You know, warm, friendly, jargon-free. I like that one. Especially as computer science teachers, there is a lot of jargon. And sometimes if we're trying to communicate with parents or students that are just learning some of that vocabulary, we want to make sure that we're doing so, you know, making it understandable.

So, yeah, and I know, too, like when I'm sending it doesn't Okay. And my screen doesn't have the Gems option. Okay. I'm going to show you what your screen might look like because so on some accounts like the free accounts, you'll actually see Gems right here. However, if you have like a school account or a paid account sometimes, let me go here. All right. So, if you're on your account, sometimes you'll have to go to the bottom where you see this the gear, if you go to the gear, you'll see Gems. So, if it doesn't show up on the left-hand side, click on the gear kind of above your name and it should show up as an option there. Yep, no problem. Perfect. Yeah, it just kind of depends. I actually pulled it up on my free personal account just because I wasn't sure if other people would would have the pro accounts but yes, you can do it on both. And then when you click on Gems and I'll do it on my free account just in case, you know, not everybody has the pro, just in case it is any different, which I don't think it is. All right. So, once you're here, and it sounds like most of us are here, I'm going to go ahead and click Gems, and we are going to scroll down to this Gem down here. Not the Gems from labs. This this other one down here. All right. And when you click on that, it's going to give you some space where you can name your Gem. So, I'm going to go ahead and I'm going to say this Gem is going to be a lesson planning coach who's going to help me plan my lessons. And then just give a brief description. Now, I have these all prepared ahead of time, so please, you're going to go at your own rate. I don't expect you to have all of this prepped right now, but you just kind of throw in a description here. So, mine just says, "I'm using it to help teachers brainstorm lesson ideas, adjust activities for different learners, create clear drafts of classroom materials while keeping the teachers in control of final decisions." All right.

Then, in these instructions, this is where you're going to kind of refer back to this purpose, audience, guardrails, tone, because I'm going to go ahead and I'm going to address all of these in that instruction category. So, let me jump back. Sorry, all this jumping. I'm going to go here in the instructions and you'll see I said, "Your lesson planning coach for K-12 teachers. Your role: Support teachers as they plan instruction. Create classroom materials and think through student needs. And then your response should be practical, teacher friendly, and easy to adapt. Do not assume that teacher wants a perfect final product. Instead provide a strong first draft that the teacher can review, revise, and personalize." Giving it a lot of kind of direction. All right. When helping with lesson planning, kind of giving it some bullet points on what I'm looking for. When creating classroom materials, a kind of bullet points of what I'm looking for here. And then when giving feedback, because we know that when we're working with AI, it's an iterative process. Like I would say 100% of the time, but like maybe maybe there's 1%, but like 99% of the time when we're talking with AI, it's not going to nail it the first time, right? You're going to go back and forth a little bit. So I also am giving it a little guidance on that back and forth. "When giving feedback, do not simply agree," baking it in. I mean, it's AI and it's deeply ingrained for that sycophancy, but maybe this might help decrease that a little bit. "All right. Point out strengths, possible gaps, and one or two practical improvements." So, I'm kind of baking it in. I don't want you to just say everything I said was wonderful. Really like work with me to make sure it fits, and then, "Ask clarifying questions only when needed," because we all know AI is just going to want to keep that conversation going. So, I'm kind of building that into here. Then I have my my guardrails: "Do not request or use student PII, right? Do not make any final decisions. Do not present generated content as verified facts. Encourage the teacher to review and adapt, right? And then I say tone: friendly, practical, encouraging, clear, not overly wordy." All right. All right. And then I'm going to leave default tool as is. And then knowledge, I'm going to go ahead and leave it as is. However, I will say that you can add files. Like for example, I worked for the school I worked at was a Google school. So everything I made was Google Docs. Like you could easily add files from your drive on here. Or if you have files like let's say you're making something that you want to help a lesson planning coach and you want to just bake in your state standards, you know, you can go ahead and upload those directly into that. So any kind of like knowledge base that you would want to bake into there that it can use to generate and work from, you can add those down here. So those are some really great options. So, let's go ahead and and then obviously when you're done, you just go ahead and hit save and then you have your Gem or your your custom chatbot.

So, you can turn to it every time you, you know, want to build a lesson or whatever your content is that you're wanting to work with. So, that's a really great tool. And then what's nice, too, is let's say like for example, you know, maybe you're I was the lone CS teacher in my school, but maybe you're not. Maybe you have some colleagues that you create a great resource and you want to share it out with your teams. You know, these custom chatbots are shareable. You can share them with other people and it can be really really great tools, not just not just for you to use, but you know, share the love, share the knowledge, right? So just a really great resource that you guys can use whether it's on your for your own design or for use to build tools for students. So we'll just go ahead and you know go ahead and build a build a Gem or or maybe you used a different tool. Maybe you didn't use Gemini, you have the Pro ChatGPT and you created your custom chatbot. What are they called for ChatGPT? I forget. I don't remember. They're called something. Oh, GPTs. Is that it? And then when Oh, I would need separate Gems for each or one that would help with all your classes. You could you could create one that would Yeah. Custom GPTs. Yep. All right. So yeah, you could definitely create multiple Gems for each class or just one overarching one and just keep the directions a little bit more vague. And then when you're utilizing the Gem, like let's say you're utilizing it for all of your classes, you could just when you're conversing with that Gem, you know, just say, "Oh, this is I'm building materials for this specific class," so it knows what file types you're wanting to reference.

Oh, so you just canceled ChatGPT Pro and you never learned how to do that. It's it's really the same way with Gemini. So, if you if you want to use Gemini free version, you're welcome to do that. You know, now that I showed you how to do it, I want to say it's a pretty similar process with ChatGPT to create those custom GPTs. Ah, I'm so glad that you're you're already you're already having a takeaway from today. That's great, right? And then if you you know if you if you've made something or maybe you don't have time to do it right now and you have an idea, but just go ahead and you know one sentence into the Zoom the Zoom chat. It kind of says, "My Gem blank for blank." You know, what does your Gem, what does this Gem that you just made do and for what audience? Or if you have an idea of one that you want to build, you just don't want to do it right this minute or you need more time or resources, you know, what is an idea you have for a Gem that you might want to make.

How does it share this Gem with my students for them to use? Okay, that's a great No, the guy. Oh, yeah. Yes. But now I have the writer's idea block. Yeah, I know. Sometimes I do that, too. All right. Let me see here. All right. So, to share, "I plan to make a Gem to help build professional development sessions for teachers and upload teacher feedback." Awesome. Yay. That's great. Building PD sessions. I use Gem for those as well. All right. So, sharing a Gem. So, let me go back to my Gem here. All right. Where do I see my Gems? Um, okay. Here we go. All right. So after you go and after you've created a Gem here, you'll see I've created two different ones, a parent email drafter and the lesson planning coach one I just made right now. If you look here, you'll see an option that says share. And then you can go ahead and change who has access to it. They'll be able to view and you can go ahead and copy that. All right. And then if you ever want to refine it or edit it, if you find that your Gem isn't working the way you want it to, you can always go back through and edit that.

All right. I want to make a Gem that can call parents and let them yell at the Gem instead of me. Oh man. Parent phone. Yeah. Can we have That's one thing. Can we have AI do our parent phone calls for us? That would be the next the next step, right? That would have AI handle that. I know companies do. I was just at a PD last week and a teacher said that she was wanted to cancel her like SiriusXM and she called to cancel it and then she was talking with an AI agent and was able to negotiate her cost down so she had a better plan for a lower price and she's like, "Yeah, I was just talking with an AI assistant," and I was like, "Wow, all right." So it is possible, they do it for the customer service at SiriusXM, maybe maybe schools are the next area here. Oh, doing doing something around sub planning. Absolutely. Yeah. Generate, you know, have some ideas baked in there for sub plans. Say, "Hey, here's what I'm thinking," and generate something quick and easy. You put in the limitations that subs have, you know, bake that into the Gem and go from there. Yeah. Yeah. And like I like this one up here a little bit. The Gem used to help students debug Java code snippets without directly giving them the answer. I think yeah, if your district allows you to use these tools with them, showing them how to use it appropriately, I think is really helpful because, you know, when I'm doing code and generating code, if I don't understand something, I turn to AI to help me understand it. So I think that that's great. You guys came up with some awesome ideas. Cuz I love it.

All right. Well, we are almost out of time. H I know, right? Well, if you thought today was fun and you thought this was a helpful 80, you know, 80 minutes of well, actually almost 85 minutes now, right? This is actually part of our Applied AI for Educators cohort, which is a full course that's all focused on AI, you know, how to understand it, how to use it, you know, how to how to build tools like custom chatbots or AI agents or, you know, AI-powered apps, all of that kind of stuff. If if that's right up your your area of interest, it's an asynchronous course, all focused on AI and how to use it and how to build with it. We have five live sessions throughout the school year and ongoing support with myself or someone from our our team here for questions and personalized coaching and all of that kind of stuff. So if that is something if you want to go further beyond what we cover we're able to cover today in this you know 90 minutes, this is definitely a really cool a really cool program that we have at CodeHS. I'm going to throw the link in the chat just so I sounds expensive. Just throw the link in the chat and, you know, that'll be the end of my my pitch here. I do I do love that course. It's one that I I I built that one and very passionate about AI. I'm sure you guys all discovered that today while I was, you know, working with you all. But I just, you know, just wanted to throw that in the chat.

And then the other one, this one is very helpful for me. This is our webinar feedback. So if you have some feedback for me, you know, anything that you liked from today, anything, you know, I could adjust for the future or just, you know, your thoughts on how our our session together went. Please, you know, take a moment or two to go ahead and click that link. It's just a Google form, not very long. You'll just select the webinar you attended which is the AI Applied AI for Educators one. And then the last link, if you are a CodeHS user, you've used CodeHS, you have a CodeHS account, if you log into CodeHS and click this link, you can you can go ahead and get a certificate for spending this 90 minutes with me by clicking that, but yeah, it is for if you you have do have to be logged into your CodeHS account. So, I know we didn't do anything on CodeHS today, but if it's easy to log in, you can go ahead and click this to get your little your certificate record that you attended this.

All right. And if you enjoyed our our webinar together, we do have a library of webinars that we keep. You can watch recordings, read transcripts, explore takeaways. Go ahead and throw so many links. I'm I'm exploding you with links right as we wrap up today. But just in case you want to revisit some of our other webinars that we've done, explore those resources. Could you hotlink the link? Are they not? The last link didn't Oh, no. Okay. Um, yeah, if you click that last one, it should have the cohort one. Oh, okay. Thank you. Sorry. Let me see. CodeHS cohorts. Yeah, this one should just take you. All right. Looks like all of them are. I can re-put the cohort one in there for you. All right. Let me go back, find that the cohort one for you. I'm going to throw it in the chat one more time just in case so we all have it. Okay. All right, it should should work. It works on my end. If it's still not working, I could give you the full the super long URL, the non-condensed CodeHS friendly version. I'll just throw that in the chat just in case that one's not working. Here you go. So, that is the the long version of the cohort one. All right. So many links. All right. So, yes, that's the cohort. Then the feedback, then the certificate. Yes, the certificate just says, "Thanks for attending. Yay, you were here," and then you'll get what it does is it'll auto-email you like a certificate.

Attendance link isn't working for you. Um if you aren't logged into CodeHS, it's not going to work. So, you will have to log in first. Yes. Yep, I can do that. It is here. I'll put them both in there. I'll put this one and then the actual here. If that shortened link doesn't work, I can put this longer one in there, too. I like to condense everything down so it's easier to digest here. And those should work. All right. Okay, we talked about our All right, if you aren't a CodeHS teacher and you'd like to learn more about CodeHS last week, I promise this is just our learn more about CodeHS. Seems like a lot of you guys use it because you were able to get that certificate. But there's our little learn more. And then of course, as we hit 4:30 on the dot, thank you guys so much for being here. I really enjoyed geeking out on AI with you all today and it sounds like you guys have some great ideas for how you're going to use these custom Gems and some ideas on how to use AI with your students and for yourself. A little bit more knowledgeable with that framework. And thank you.