Event Replay: K12 Showcase: Lessons From Teachers and Administrators

[00:00:50] Mark Murray: Hi everyone. Thank you for joining us. Welcome to the OpenAI Forum. I'm Mark Murray, the Editorial Director at OpenAI. And as districts across the country head back to school, we wanted to highlight how system leaders and educators are using AI to support teaching and learning. Brian Johnsrud, OpenAI's Global Head of Educational Engagement, will first facilitate a conversation with district leaders about how they're using AI in their districts. Then, 4th grade teacher Francesca Summers will share how she uses ChatGPT to create standards-aligned, differentiated reading lessons. We'll leave time for audience questions at the end and I'll pass it over to Brian. Brian?

[00:01:41] Brian Johnsrud: Thanks, Mark, and hello, everyone. Before I came to OpenAI, I was a K-12 teacher and then a college instructor. So, when I think about new technology in education, I don't start with the technology. I start with expert teachers and school leaders and what their goals are for the classroom and how technology's going to help fulfill those goals. Educators know better than anyone when a student needs a different explanation or a personalized resource or support. They know what's going to keep students engaged and actually excited about learning. Who takes a standard written on a page and turns it into that dynamic learning moment that changes what a child believes that they can do. That's why this moment feels so full of possibility and I'm so excited to spend time with the innovative educators and leaders joining us today. AI has quickly moved beyond answering individual questions in chatbots. New agentic tools, connectors, plugins, and automations can now work across the curriculum materials and specific course context a teacher chooses and they connect to their other essential tools and help carry out new multi-step pedagogically informed work taking an idea to something ready for the classroom. That could mean turning one lesson into several pathways for different learners, building interactive activity, or spending less time assembling materials and more time with students. It's an exciting moment in time and a journey where we're proud to be co-designers of how far student potential could go and in that journey it's the educators who first draw the map they choose the destination what students should understand where their starting points are and what meaningful learning looks like and what's possible. The researchers and pedagogical experts we work with they help measure and test the roads on that map validating what approaches are most effective for which students and in which settings and our job here at OpenAI is to help build bridges that can unlock new routes and when we get this right together schools can take students to places that would have been difficult or even impossible to reach before. So our goal is simple to build with educators learn from educators and empower educators and that's what today is all about. The innovation we're celebrating isn't happening somewhere far away it's already happening in districts and classrooms in the hands of the leaders and teachers you'll hear from today. Patrick and Gautam thank you for joining us let's get started. Can you please introduce yourselves briefly maybe starting with you Patrick.

[00:04:12] Patrick Gittisriboongul: Awesome, again thank you Brian thank you to the team at OpenAI for having me on this panel. My name is Patrick Gittisriboongul and I'm the proud superintendent of Lynwood Unified School District in Los Angeles County. We serve approximately 10,000 students across 12 elementary to middle and to high schools. Lynwood Unified is a very diverse urban school district in southeast LA. As superintendent, you know our work is really grounded in a simple mission is making sure that every student has access to high quality education, strong support systems and the opportunity to succeed. I've had the opportunity to serve in a number of different roles in education. I've served as a classroom teacher, taught math about 26 years ago so I'm dating myself now, but also was in Ed Tech, assessment and accountability, data and district administration.

[00:04:58] Patrick Gittisriboongul: And district administration. And now as superintendent, I have the opportunity to look at education from a systems and organizational perspective. I think I would say my path to education is also somewhat unique. Before becoming a teacher, I worked in the private sector as a web developer during the dot com era, the dot-bust era. I eventually transitioned into education, became a math teacher. And so that experience has really shaped how I think about technology. Technology is only valuable when it ultimately improves the experience of people using it. And so from the classroom, I moved into ed tech, leadership roles where I had the opportunity to work on instructional technology, assessment, accountability, data, and innovation across school districts and county systems as well. So essentially I've kind of spent my career sitting at the intersection of technology, teaching and learning, data, and organizational leadership. Now as superintendent, I think my responsibility, again, is very much, much more broader. It's about leading the entire system, really supporting our schools and our educators working with our board, partnering with our families and communities, managing resources responsibly, and most importantly staying focused on students. And I think that's important for the conversation today. And I'm very happy to be here. So thank you very much for having me.

[00:06:11] Brian Johnsrud: Thanks, Patrick. Yeah. Gautam, can you introduce yourself?

[00:06:14] Gautam Sethi: Sure, absolutely. Good afternoon, everyone. My name is Gautam Sethi. I serve as the chief information technology officer at Fairfax County Public Schools in Northern Virginia. We are the ninth largest school district in the nation, serving approximately 180,000 students in 200 schools with about 40,000 staff supporting them. We have a diverse population, about 194 different languages spoken in Fairfax County Public Schools. So it is a large, complex school division where the needs continually outstrip the resources. And we start working with tools like AI. We are excited about the potential and possibilities of supporting the classroom teachers, supporting the back office operations, and helping build towards that stronger, tighter teacher-student collaboration in the future. Unlike Patrick, but somewhat similar, I did not follow his pathway into the classroom and to technology. I actually just stumbled into it at New York City when I joined there as a developer working on a project to modernize their student information system. And since then I've been sort of involved by the challenges in education and how technology can support it. 20 years, four school divisions later, I'm still at it but excited with the new technology because finally the promise of technology supporting education I think is getting closer and real. Thanks so much.

[00:07:46] Brian Johnsrud: Wow, wealth of experience and expertise between the two of you. Let's dive in. Patrick, my first question is for you. So every school year it seems like there's a different theme or vibe or motto that sort of can encapsulate how AI is being integrated that year and how teachers, parents, students, and school leaders yourself are thinking about it. In your school community, what would you call this year's theme or motto about how you're thinking about AI?

[00:08:15] Patrick Gittisriboongul: Thank you for the question. I would describe this year as really moving from experimentation to more integration. And I think it's because of how quickly the conversation and these platforms have evolved. A few years ago, I think the conversation was mostly about are students gonna cheat, should we block ChatGPT, what are teachers supposed to be doing about this? I think now we're entering this phase of society and even in schools where everyone is experimenting with these platforms. And so people start asking now, how do I write a better prompt? Can AI create a lesson plan? Can I draft an email? Can I summarize a document? And so that experimentation is really important because it helps build that familiarity that many of our teachers and our staff and our community and families have. So I think we're entering that different phase. So now that question is becoming less, what can AI generate for me? And more, what can we do differently because it exists? And so I think that's a much more interesting question. I think for teachers and staff, I think that could really move beyond asking AI to create a worksheet and instead using it to help analyze student work, identify misconceptions, differentiate instruction, think about the next instructional move. I think for administrators and district leaders, I think it really means moving beyond drafting this parent letter and instead helping synthesize attendance, achievement, school improvement and other information to really identify where leadership attention is needed.

[00:09:56] Patrick Gittisriboongul: Where leadership attention is needed. And so at the district level, we're beginning to look at entire workflows. So things like, for me, how do I prepare for board meetings? How do we review student outcomes? How do we develop executive briefs? How do we communicate complex information? As a school district leader, oftentimes the data that we have, the information that we provide can be very complex for our community and for our parents and our students. And so helping communicate that out, how we preserve that institutional knowledge that we have in an organization is very important. And so I think when it comes to that theme, I think we're moving away from AI literacy and more towards AI fluency. And so that literacy is really understanding what AI is; I think fluency is really knowing when to use it, how to use it, and then how to evaluate with it. And so I think that's where our educators, our leaders, our administrators are really increasingly needing to be. And so we're really moving from asking what can AI generate to now what can we do differently because it exists?

[00:10:56] Brian Johnsrud: Yeah, it makes a lot of sense. Even hearing you talk, the verbs that shifted, like drafting and understanding to synthesizing, preparing, communicating. Gautam, that leads me to a question I have for you. One shift we've seen, especially over the past year, is dramatically greater capabilities for AI to go beyond chat and simple content generation. Now we have things like automations and plugins and agents and use cases that weren't possible last year. How are you thinking about those changes in the technology?

[00:11:31] Gautam Sethi: Yeah, Brian, super excited about this one, right? As a technologist, and even on a personal level, I said I got into this, into the world of education as a technologist helping education improve a system. I was trained as a programmer, undergraduate in programming and computer science. And almost invariably you end up stopped kind of working on that technology as you work through the organization charts. And in the last, I would say six months, the availability of tools like Codex and all the other agentic tools that are available in the market allow anyone with an idea to be able to get back in and at least prototype that idea. I have programmed more in the last six months than I say I would have done in the last 10 years purely because I can go have that meaningful conversation where my idea can be materialized into a working prototype with just that agentic conversation. And these are really powerful tools, right? So as a technologist, I can evaluate the power of these tools. As a user, I can see the value of these tools. And when you combine those two things, the unlocks are just absolutely, there's no limit to it, right? Because you can have somebody with an idea whether that's in a transportation department and how to optimize traffic flows or parent request in finance or in the classroom where a teacher is thinking, how do I have a different way of getting to that student, right, that one unique student.

[00:12:59] Gautam Sethi: Those ideas in the past would have been a long conversation. I got to have to go talk to a technologist. I have to go engage with the central office team, hire a consultant, buy a product, any number of things. Now you can go from an idea concept to a prototype to possibly validating and thinking and saying, well, that worked but, or, and. It just helps you sharpen your thinking so much more quickly. And that was entirely impossible for the last, I would say, two decades, where we all wanted technology to serve the purpose of education. But getting you to do it meaningfully and quickly was just so hard and so time-consuming or costly on resources. Anybody with access, and now with ChatGPT for Teachers, you have access to tools like Work or Codex for that matter. And you can use those tools to be able to take your ideas to some version of reality and understand if that is meaningful. It's gonna be helpful if it's useful. And we have had real impact. Our staff has been building tools. We had the Jam session, which was so powerful and so engaging. Teachers just love that. Even the school staff and the central office staff that was there. Because you could take an idea in the course of, I think we were there for six hours. And by the end of the day, you could materialize it into a working, meaningful, tangible output. I think that's really, really powerful for folks who haven't yet tried it. One simple piece of advice, just build, start building. And once you do that, it just unlocks so many opportunities and potential possibilities.

[00:14:31] Patrick Gittisriboongul: Brian, you might have to add a little bit to this question as well. So I think I like that response, Gautam. I think when it comes to skills agents that Codex, ChatGPT and ChatGPT Work has really kind of elevated, I'm starting to see now the possibility of moving to change workflows. And so that's significant because, as I mentioned earlier, school systems can be very complex.

[00:14:54] Patrick Gittisriboongul: Be very complex. And so school districts, organizations, they're a collection of thousands of workflows. As an example, attendance intervention is a workflow. Even as superintendent, I have to prepare a board agenda every week. I got a board meeting coming up this Thursday. So board preparation is a workflow. Hiring is a workflow, purchasing is a workflow. All the aspects within an organization are very important for us. And so how can OpenAI, ChatGPT help us execute these workflows more consistently and intelligently? That's an experimentation that we're starting to develop internally as well. And so I got to mention earlier about building platforms, building tools, building proof of concepts, if you will. And so I do think that this new wave of being able to develop like workflow system automation is just unprecedented.

[00:15:51] Brian Johnsrud: Absolutely. It strikes me the examples you're showing, Gautam, are so innovative. And I think for many years, teachers and schools are incredibly innovative. And so much of that innovation has come from constraints and limited resources, which actually can very much help with innovation. And when some of those constraints are no longer there because they have access to code or to build or to manage data, this sort of outpouring of innovation that can come from that is really exciting. Patrick, I'm wondering, as these tools and technology advance so quickly, we face potentially the risk of those who are following along and are learning the most up-to-date recent capacities, and those who are maybe using the tools for what they were capable of a couple of years ago. So my question for you is, how do we best build the capacity for teachers and administrators to learn about and then adopt those more advanced or powerful workflows so that they can fully utilize what's possible, both in their instruction and for, in general, improving outcomes for students?

[00:17:04] Patrick Gittisriboongul: Yeah, I do think that we have to change the way that we approach professional learning, especially when it comes to AI. I do think that being in public education now, 25, 26 years before coming into it, I came from the dot-com era, so I know what traditional software training looks like. And so if our PD is essentially, here are our 10 prompts, here are the buttons in this product, we're preparing really people for yesterday's technology. Instead, I really think, I mentioned earlier about workflow. I do think that we need to teach people to deconstruct their work. So things like take something that you do repeatedly and ask what triggers this work, what information do I need, where does this information come from, what steps do I repeatedly perform, where do I spend unnecessary time, what decisions need to be made, what could AI appropriately assist with? I think that's how you, and how I think we begin thinking about workflows rather than prompts. I know the industry has talked about loops, giving more things, giving more context, context engineering, and so as an example, instead of giving a principal a three-hour workshop introducing AI agents, I would rather say bring one recurring problem that takes too much of your time. Maybe it's analyzing attendance, maybe it's reviewing school improvement data, maybe it's communicating with families, and then take that actual work and then redesign what that workflow looks like. I think the same thing can apply to teachers. So we don't want to teach a teacher how to ask AI to generate a lesson even though I know that's also valuable, but maybe we can look at student work. Maybe we can look at identifying patterns, misconceptions, have the teacher evaluate, synthesize that analysis, and then determine the grouping supports differentiated instruction, the type of scaffolding needs that students need. And so I do think that we want to ask teachers, you want to make sure humans are in the loop. We can ask these questions about, what does ChatGPT see? What did you see? Where do you disagree? And so that builds that professional judgment, that builds that human in the loop, but also builds the type of AI fluency that we want. I do think that's very critical because the goal isn't to remove professional judgment. The goal is to increase the capacity of the professionals in the room exercising that judgment. So we don't want to teach teachers and staff how to use AI. We want to teach them how to redesign to work with AI.

[00:19:52] Gautam Sethi: If I...

[00:19:55] Brian Johnsrud: Please, Gautam.

[00:19:56] Gautam Sethi: Yeah, Brian, one quick note right after what Patrick was talking about, right? The way we have seen some of the initial sort of success or the ability for people to kind of start using the tools has been trying to find a problem which is close to them, right? Almost everyone I've seen have that aha moment, came to the tool and said, I wish X could happen. I wish somebody could go update my calendar with all of those events. I wish, whatever that wish was, and then we would be like, all right, let's try this out. Why don't you just say what you wish to this genie, that machine, and see what it produces. And you know, eight times out of ten, and it depends on how you frame your wish, right? It's not as simple as I wish, and it's done. It's I wish, and then you give it a list of clear context, clear outlines, and what you hope success looks like. And almost always, especially the new and modern Sol is just absolutely amazing. You'll end up producing a workable output or value, and I think when people see that, the first time I've seen anybody see that, it's almost like if I had been in the classroom, I can imagine the teacher experience when they see the kids' eyes light up when they get the concept. That's how I see adults or professionals, teachers, others, when they use the tool, and the first time they get that click, it's like, oh, my God. And after that, they're on their own journeys, right? Then they're discovering the tools and doing things and nothing's stopping them other than the credits, which we have to figure out.

[00:21:29] Brian Johnsrud: Yeah. I love that. I think in education and across industries, we're all in this funny moment where I think the challenge now is changing our perspective of what's possible. To be asking the questions, well, what if I could do this? We haven't been asking because it just wasn't possible before to have that sort of ah-ha moment. Patrick, I like what you mentioned about, really, it's becoming metacognitive with ourselves and asking, how do I perform these tasks? Who better to actually be reflexive about that than educators, because that's what we teach our students, show your work, practice metacognition, and be sort of aware of your thinking. So it's an exciting time. In the spirit of experimentation and innovation, Gautam, what's something you've tried or you piloted last year that was effective and that you're now thinking about implementing in a bigger way this year?

[00:22:28] Gautam Sethi: Wow. One thing. Okay. Let me try to narrow it down. We wanted to sort of create this meta-collection of things where we can catch people's ideas of how I wish it could do X, right? So we created an internal tool called Idea Catcher. This goes after Idea Catchers, if you've seen them in real life. And what that essentially is, somebody can come in and engage, and there's a little bit of AI there where they can say, I wish I could get X, Y, Z done. Or, this is a problem that I see. Or, recruitment takes X amount of days. I wish it could be done any faster. So they can come back with a raw idea. And then there's a little LLM agent which processes that idea with them, so it kind of goes back and forth with them to say, Hey, what would you want to see? How would you want to see it? How much time would it save? What do you estimate those things are? And at some point it says, This is great. Are you ready to record this idea? And then it kind of captures the idea. So we have a catalog of ideas where somebody says, I wish this could happen. And in a division like ours, those catalogs fill up really, really fast. I think we have a backlog for IT now. There's about 200 things sitting in there for us to do. Which then begs the question, Who starts building those ideas? So there's two different concepts of competing. We can train all the staff who are in the field, kind of building themselves because they all have access with ChatGPT for teachers. These tools are generally deployed. We would still figure out sort of the governance of structures to manage data in between. We're also kind of taking a shot at the industry from an AI industry perspective on a forward deployed engineer. We made our own versions since we can't afford an FTE in school districts. That would be more than a superintendent's salary. So we are going with a forward deployed intern, if you may, right? This is the resource-constrained version of education kind of building. So we're working with bringing college interns. And we believe the tools are so powerful that you can bring in an intern within context of a business problem, give them a tool, and ask them to start shaping and thinking that idea right. I can go through a governance pipeline where somebody can decide if this is ready for production in real life. But you can do that initial experimentation. So our idea catcher is going to meet forward deployed interns this year. And hopefully by the end of this year, we'll have many of those materialized and delivered.

[00:24:50] Brian Johnsrud: Wow. What an exciting project for a college intern to work on.

[00:24:56] Patrick Gittisriboongul: Brian, you mind if I jump in here as well?

[00:24:56] Brian Johnsrud: I was just gonna say, Patrick, give me an example.

[00:25:01] Patrick Gittisriboongul: Yeah, so I think to Gautam's point, I think one of the things that we're also experimenting with is really what I call this kind of executive operating system for the organization. So, essentially, I think one of the things I'm beginning to start to do is to codify how the superintendent... I've been in this job like a year, so I'm entering my second year as a superintendent, and so now I'm thinking about how superintendent executive leadership, how principals, how the adults, we can get work done across areas like strategy, board governance, student outcomes, you know, budget, communications, labor, attendance, as I mentioned earlier, accountability, but all the district functions that we have across the organization.

Layering that AI, you know, and ChatGPT, and work, and Codex into that operating model, I think those are the things that we're thinking of. For example, you know, I think the idea of exploring this role-based AI support for superintendent work, whether it's a cabinet advisor, whether it's fiscal considerations, whether it's legal issues, I think that that's an area that we're starting to really, you know, automate and kind of figure out how those workflows and instructions can be established. And so, you know, I have this readme markdown document that really helps give me recipes, if you will, across the organization on really how we, you know, create that institutional organizational knowledge and context that matters.

And so that area around decision memory, that area around rationale, commitments, timelines, follow-through. And so codifying that and then making sure that good practice, effective workflows, is established, those appropriate guardrails that we put around it. Those are the things that I'm really, really excited about. So, you know, I got a lot of markdown files in my role. So if you don't know what markdown files are, they're recipes. I just got them. You know that I'm a former CTO.

[00:27:09] Gautam Sethi: It warms my heart to know that a superintendent can talk about the markdown files. We are certainly moving in the right direction, as a technology for sure.

[00:27:19] Patrick Gittisriboongul: Yeah, absolutely. So I've been AI-filled. I just think that it really helps with some of the work that we have to do.

[00:27:26] Brian Johnsrud: I love that this is going to be the new catchphrase between schools: "What's the coolest markdown file you built this week?" They're all chasing the best superintendent.md file on GitHub. That will be the answer.

[00:27:41] Brian Johnsrud: Amazing. Well, thank you, thank you so much to both of you for sharing your insights. As I was saying earlier, this is the state of where we're really at in AI. The limits are only our imagination. And so sharing what you've been working on as really innovative leaders at the cutting edge is so valuable. We'll come back to you both for audience Q&A later.

[00:28:09] Brian Johnsrud: Now I'd like to bring on Francesca Summers, fourth grade ELA and social studies teacher, to demonstrate how she uses ChatGPT to build differentiated lessons. So excited to have you here, Francesca. My colleagues first met Francesca at an OpenAI Academy Jam this summer. So OpenAI Academy and the Walton Family Foundation hosted eight hands-on jams around the country to help K-12 teachers and administrators really dive in and learn practical ways to use ChatGPT in their work.

[00:28:48] Brian Johnsrud: Welcome, Francesca. Can you please start by introducing yourself and sharing a little bit more about what you'll be showing us today?

[00:28:52] Francesca Summers: Yes, certainly. Thank you, Brian, for having me. Once again, my name is Francesca Summers. I teach in Clayton County Public Schools. I had the pleasure of being at OpenAI this summer where the facilitators and everybody was were great. I was able to interact with the program and really enjoyed the interactive lessons that they had for us to do. So today I will be explaining to you how I use OpenAI in my fourth grade writing classroom.

I don't start with AI, I start with the standard, what I need students to learn and what I'm seeing in their writing. For me, AI is a tool that supports my teaching. It doesn't determine my instruction, so let me show you what it actually looks like in my classroom. Before I bring OpenAI into the picture, I want my students to know what they're working toward.

[00:29:48] Francesca Summers: Towards what are my students learning? So I take a look at the standard. I break it down into a success criteria that my students and I can actually see in their writing. We can look at a piece of writing and say, do I have a clear opinion? Did I give good reasons? Did I support those reasons with evidence? Did I connect my ideas? Did I bring my thinking to a conclusion? That gives us something concrete to look for. When we're looking at their writing, we're not just saying that doesn't sound right. We're able to identify exactly what they're working toward and where they need support. And this is where I look at student work. One might have a strong opinion but need help organizing their ideas. Another might have good reasons but their evidence doesn't really support them. And another might have all the pieces to be ready but they need more depth. So my next question becomes, what does the student need next? And that next question drives my instructional move.

[00:30:47] Francesca Summers: This is where I bring in OpenAI into the process. I'm not asking ChatGPT to tell me what to teach. I've already made that decision. I know the standard, the learning target, the success criteria, and what I'm seeing in my students' writing. I'm asking it to help me think through possible supports for students who are working towards the same goal. That's an important distinction for me. AI enters after I've identified the instructional need. The standard stays the same, the goal stays the same, the support changes. Same standard, same goal, different support.

[00:31:33] Francesca Summers: Now this is where AI becomes really useful. I know my students. I know we've already taught what they're struggling with and what they can handle. So I look at what ChatGPT gives me and think, what actually makes sense for the student? Maybe it's a sentence stem, a graphic organizer or a word bank, or maybe a challenge question for a student who's ready to go deeper. And that's the part I want to emphasize. AI gives me options. I make the decision, the instructional decision. I review it, I select it, I adapt it, and then I teach it. So when I step back, this is my actual workflow. I start with the standard. I establish the learning target. What does successful learning look like? The success criteria. Student writing, what does each student need next? I identify their need. I use OpenAI for support. And how can OpenAI help me generate the best possible supports for my students? I'm either going to select it, adapt it, or reject it. And then I look at the most important part. Did that support move the learning forward? For me, that's what using AI responsibly looks like. The standard tells me where we're going. The success criteria tells me what I'm looking for. My students' work tells me what they need. I know where I'm going. I know what my students need. And AI helps me think through how to get them there. So for me, it's pedagogy first, AI second.

[00:33:05] Mark Murray: Francesca, thank you so much for that great demonstration that you just gave us, and thanks to all of our guests. Now we're going to actually take some questions that they have for our guests from our great audience. The first question we ended up getting from our audience is from Daniel Green, and he asked, what are districts seeing in terms of parent sentiment towards AI in classrooms? Are some parents actively opposed and what concerns are driving that?

[00:33:35] Mark Murray: Patrick, I want to throw that question to you, and I even have another question. What are you doing to alleviate those kinds of concerns coming from parents?

[00:33:40] Patrick Gittisriboongul: I see both sides. I think there's, you know, being pro-AI or anti-AI. I do think that it's really more nuanced than that. I would argue that most parents aren't necessarily opposed to AI itself. I think they want to understand what my child is doing with it, why are they using it, what information is being shared, and is this actually helping them learn? And so those are reasonable questions. I think some of the anxiety that parents or concern that I have two kids myself. I got a 21-year-old and an 18-year-old, and so seeing AI through the lens of ChatGPT, Codex, and then immediately thinking that something is cheating, that students having AI do the work for them, I think there's always a concern about privacy, inappropriate content, misinformation, and whether students will become overly dependent, you know, that cognitive offloading. I think the mistake that districts and organizations make is that they don't engage with parents early enough. It's crucial to bring them into the conversation from the get-go, explaining not just the mechanics of AI but the educational value it can bring when used responsibly. Transparency is key here. We need to outline clearly how AI is integrated into lesson plans, how it's monitored, and what measures are in place to protect student data. When we take the time to show how AI can enhance learning outcomes and equip their children with future-ready skills, many concerns are alleviated. For instance, using AI to personalize learning allows for more targeted support, which parents generally appreciate once they're aware of the benefits. It's about building trust through communication and education.

[00:34:46] Patrick Gittisriboongul: Saying that, you know, don't worry, we have an AI policy. To me as a superintendent, I don't think that's enough. And so I think we have to be able to explain what the instructional purpose is. If I'm a parent, which I am, I should be able to understand the difference between AI doing my child's thinking and AI helping my child become a better thinker. And so for example, you know, instead of saying, write my essay, you know, have a student write an argument and then ask AI, challenge my reasoning. You know, what evidence am I missing? One of the things that we rolled out here in Lynwood Unified School District is a district-wide instructional focus on claims, evidence and reasoning.

[00:35:33] Patrick Gittisriboongul: And so that aspect, like if you say that to many teachers, they'll say, I know what CER is, I can adapt it. Now, applying AI enabled learning, support learning, what does that look like when the tools, if students are allowed to use it, what does that responsible developmentally appropriate AI-supported learning look like? And so I don't think parents and the community, we can't just need an AI policy. I think they need to understand what the instructional purpose is. And, you know, in my experience, it's always around this trifecta of curriculum, instruction and assessment. And so all of those three aspects need to really be aligned. I don't know if that answers your question, but that's how I would respond.

[00:36:15] Mark Murray: I think you did pretty well, Patrick. Gautam, the next question is yours. And this comes from Lana Romanova who asks, when AI saves teachers time, what evidence would you do to convince that it's also improving student learning, not just making existing workflows faster?

[00:36:32] Gautam Sethi: Yeah, I think there's two parts of this, right? And everyone's trying to figure out how AI gets close to the student. In our opinion or at least the work we are doing, we're focusing heavily on adult AI literacy. This technology came so fast. Everyone, including the technologists, are still trying to catch up and keep up with the technology. And having that technology be well understood before it can be well distributed, I think is the path that we should all be looking for. So the value of AI improving student outcome is a long horizon conversation, which should be researched, and I know there's a lot of research institutions working towards it, and we continue to collaborate with them to see where that ends up.

[00:37:14] Gautam Sethi: But in the near term, if we can just save our teachers 30 extra minutes a day by removing some of that administrative overhead, and that is very personal. I don't think we can say, there's this one magical thing we did that every teacher now has 30 extra minutes. Every teacher has a unique situation, their own workflows. If we can teach them, train them, get them to be able to use these tools to save themselves time, we feel that time gets reallocated towards face time with students, which ultimately everybody knows by every research that has happened over the last 50 decades or 50 years that teachers with students in face time will produce better outcomes for students all day, every day. No technology can ever substitute that.

[00:37:59] Patrick Gittisriboongul: Yeah, can I add to that as well? I do think that efficiency is very valuable. I think that teacher time is incredibly valuable, but ultimately as a superintendent, I really wanna make sure I connect efficiency as something that changes the lives of kids, right? So I would look at this kind of chain of evidence, this listing of improvements. So if AI saved the teacher two hours analyzing student work, what happens with those two hours? Again, part of our vision 2030 is to make sure that positive developmental relationships is a primary focus of us as an organization. Like teachers developing relationships with kids, making sure that there's a caring adult at the end of a classroom, working with kids one-on-one in small groups. I love that. I think instruction needs to change. Students are receiving more feedback, interventions happening sooner, students getting more opportunities for revision. And so that experience, that student experience, to me, is what matters tremendously. So, kids want to feel safe, welcome. Like they know who they are, their name, their strength, their needs, all those things play a part in that classroom. And so that's how I would address that.

[00:39:20] Mark Murray: Thanks Patrick. Francesca, you end up getting our last question. This comes from Jason DeLuca. And on a similar line asks, if you have five minutes to convince a fellow teacher who's kind of AI reluctant, what ChatGPT workflow would you demonstrate to make like demonstrate the value immediately in both time saved and in quality?

[00:39:43] Francesca Summers: Time saved and quality would be.

[00:39:44] Francesca Summers: Time saved and quality would be scheduling, pull outs, small group instruction, and your own dynamics of teaching what the curriculum is asking you to do. I personally had to do this because I was confused. I have three to four groups being pulled at different times all within 45 minutes of each other. During that time frame, I also have to conduct small groups with the students that are left in my classroom, and then the students that come back to my classroom. So for me, I had to visually wrap my head around it because it's a lot to endure when you have a classroom with 25 students, and it's like a Walmart. So what it enabled me to do was to actually look at my schedule, who's being pulled out, what their need is, and then the students that were left remaining. So it helped me organize my small group, what I would teach them for that five days, because I put it in for five days, what I would reteach the students that came back into my classroom to get them up to par with the fourth grade instruction. So not only did it save me time, it helped me organize my thoughts, gather the students, pull the resources that I was asking it to pull for me because they are on different levels. So differentiation is a key. With ChatGPT, OpenAI, any teacher, I would sit down with them and walk them through the steps because you're going to save time. You're not working outside your contract hours, which some of us enjoy. And it's a bridge. It's an amazing bridge to see what can become and what is available. Because I've become an expert. Well, not an expert, but I have whispered into several people's ears in my building, and what they've done this school year with it has been amazing.

[00:41:39] Mark Murray: I have to go back and ask for some pointers. Keep whispering into those ears. Francesca, Patrick, Gautam, thank you so much. We really appreciate you joining us.

[00:41:52] Mark Murray: And if you want more resources on how to work with AI, I recommend checking out the OpenAI Academy, which has trainings tailored for K through 12. And we're going to drop a link into the chat on that. Thanks again for joining us. Until the next time.