Featured Speakers:
- Dr. Ann Edwards, Senior Director, Mathematics Education, WestEd
- Sarah Nielsen, Research Associate, Mathematics Team, WestEd
- Dr. Drew Nucci, Research Associate, Mathematics Team, WestEd
Host:
- Danny Torres, Associate Director of Events and Digital Media, WestEd
Danny Torres:
Hello, everyone, and welcome to the 20th session of our Leading Together series. In these 30 minute learning webinars, WestEd experts are sharing research and evidence-based practices that help bridge opportunity gaps, support positive outcomes for children and adults, and help build thriving communities. Today’s topic, How to Support STEM Teachers in Developing Generative AI to Improve Teaching. Our featured speakers today are Dr. Ann Edwards, senior director of our mathematics education team at WestEd, Sarah Nielsen, research associate for our mathematics team, and Dr. Drew Nucci, research associate for our mathematics team.
Thank you all very much for joining us. My name is Danny Torres. I’m associate director of events and digital media for WestEd. I’ll be your host. Now, before we move into the contents of today’s webinar, I’d like to take a brief moment to introduce WestEd. As a nonpartisan research, development, and service agency, WestEd works to promote excellence, improve learning, and increase opportunity for children, youth, and adults. Our staff partner with policymakers, district leaders, school leaders, communities, and others, providing a broad range of tailored services, including research and evaluation, professional learning, technical assistance, and policy guidance. We work to generate knowledge and apply evidence and expertise to improve policies, systems, and practices. Now I’d like to pass the mic over to Ann. Ann, take it away.
Ann Edwards:
Thank you so much, Danny, and thanks to you all for joining us today. We are really excited to share some insights from our recent research into how STEM teachers use generative artificial intelligence in their practice and how they can be supported to deploy GenAI to enhance their instruction. With this work, we seek to contribute to the rapidly expanding conversation about GenAI and education by focusing in on the perspectives and needs of STEM teachers and how the context of their work shape how they use GenAI.
But before we get started, we’d like to hear your thoughts about GenAI in STEM teaching and learning. We’ve created a Slido, inviting you to share one way that you think or hope that GenAI could improve STEM teaching. To get to that Slido, you can use this QR code on your phone, or there’s a URL provided there. So we’re gonna give you a few seconds to head over to the Slido, and then we’re gonna display what you all are saying about what you think GenAI could do to improve STEM teaching.
Okay. Oh, here we go. All right. So I’m gonna try to see if I can read this. I apologize. It’s very small. So folks are saying virtual tutoring, creating interdisciplinary tasks or real world, right? Differentiating problems, customized learning, more engaging assessment tasks. So I’m seeing themes around personalization, customization, application. Using AI to be able to provide greater supports for particularly when their teacher is not available or when on their own. Differentiating.
So what I’m seeing here is applications that are really student focused in terms of creating different kinds of learning opportunities for students, whether that’s more personalized, customized, application oriented, and then also ways to use AI that support teachers to be more responsive, to get some of their work done more efficiently, to enhance their assessment practices. And can you scroll back up, Danny, to the very toppy top? Yes. And overall, I think improve how teachers prepare to work with their students in more effective ways. All right. Thank you for sharing that.
When we head back to the slides, I think that what you’ll see today in what it is we’re sharing is really aligned with the kinds of things that you just shared in the Slido. And what we’re gonna be sharing today are findings from a couple of studies that were conducted as part of the AmplifyGAIN AI Research and Development Center. That center is led by our colleagues at the University of Washington, and AmplifyGAIN is one of the AI R&D centers funded by the Institute for Education Sciences at the Department of Education. And just real quick, if you’d like to learn more about the overarching center and its work, we’ve provided the link to the AmplifyGAIN Center in the chat.
So today we’re gonna dig into some of those research studies that we’ve conducted in the first year, provide examples of what we’re calling substitutive, amplified, and transformative use of GenAI, and then also talk about supports teachers need for engaging in transformative student-centered use, some of which, of course, that you already articulated in the Slido. Okay.
So a little bit more about the studies before I hand it over to Drew and Sarah. The studies we are reporting on here are the first exploratory studies of the center and were conducted in our first year. Further studies will be building upon this work in years two and three, and we’ll share that in the upcoming years with you. In just a minute, Drew will first tell us about a nationwide survey of approximately 1,000 K-12 math and science teachers who are on the RAND teacher panel. The survey asks teachers about their GenAI adoption and use, the supports that they have experienced, and their perceptions of what they need to use GenAI effectively in their instruction. The survey was fielded in May of this year, just a few months ago, and we released the report just this month. The QR code here on this slide will take you to the full report.
Then Drew is going to hand it over to Sarah to report on an interview study that sought to dive more deeply into math and science teachers’ actual uses of GenAI in their teaching, as well as the aspects of their social and institutional context that shaped their use. In those interview studies, we conducted 90 minute interviews with about 15 teachers who represented a range of AI adoption. And then we spoke for 60 minutes with a few of each of their professional colleagues and collaborators, and to dig into the context of use more deeply. And now here is Drew to tell us about the survey study.
Drew Nucci:
Thanks, Ann. Lovely to be with you today. Thank you all for coming. So I’m gonna be up at 30,000 feet, really looking at large groups of teachers nationwide and the survey results, trying to unpack what are their uses for generative AI in their instruction, and then how do they experience their barriers and supports for adoption? So like Ann said, the report is very, is huge. So, you know, take a look. I’m just gonna focus on a couple things.
This first bar chart here, we ask teachers to comment on how much they’re using GenAI, and I just wanted to draw your attention to a couple things. One is the top bar. This 35% represents people who’ve tried GenAI and want to use it more. And then the next three bars, all those bright purple bars there, that represents people who aren’t using GenAI very much. Either, they’ve never heard of it, they’ve used it, they’ve tried it and they choose not to use it, or they’ve heard of the tools and they haven’t used them yet. So you put all that together, that’s about 85% of the math and science teachers saying they have really light use.
And so there’s a fertile ground for support here. On this second slide, I’m gonna just dig into for the people who said that they did use generative AI for instructional tasks, what were they doing? So you could see here this bar chart, the top two, 76%, 61% is really about creating materials, lesson planning, assignments, assessment creation. 50% said they use it to support in class instruction, meaning they use it in class to be more responsive to students who learning trajectories might be other than they anticipated. I wanna draw your attention to the 32% and a few bars down, the 21%, who are using generative AI for differentiating for students with disabilities or ELL supports. We would love to know what that entails. And so we are hoping to do more research on that.
21% use it for professional learning. In our interviews, that looked a lot like content, especially for science teachers. Some may be a chemistry major, for example, and be tasked with teaching earth science and need to understand concepts better. 13% said they use it for assessment and grading. In our interview series, principals expressed a little bit of hesitancy about that because they were worried teachers were putting student data into unprotected systems, which has policy implications. And you can see 10% of the teachers are teaching about AI. And one of the things we keep saying is, we shouldn’t be teaching with AI without teaching about AI. And so 10% of the teachers already starting on that important work.
If we look then at what supports the teachers are getting, just some big numbers stand out. Only 22% of the teachers have received formal professional learning on artificial intelligence. Only 5% are in institutional contexts where they have school and district formalized AI use policies. So there’s definite grounds for really thinking about support. And that support’s important because this bar graph shows us what do the teachers say are the key barriers? And like any innovation, the top bar, 61%, is the time we’re required to learn how to use it. But that’s related to the second bar and the fifth bar, where 54% and 38% respectively say there’s insufficient training opportunities, insufficient professional development, and then 45% saying there’s unclear district guidelines or policies.
Interestingly, on the bottom of this graph, you can see that things like GenAI quality output, technological infrastructure, or technical difficulties with tools are not cited as much as barriers. So what we’d like to do now, Danny’s gonna start a poll for us. And what we’d love to do is just hear from the people in the room, you know, does your district have an AI use policy either for teachers or for students? And you can say yes for teachers, yes for students, or both, or neither, or you’re not sure. Does your school or district have a set of approved AI tools they endorse? And we got the choices yes for teachers, yes for students, yes for both, no, or you’re not sure. And has your school or district provided professional learning for teachers? This is a a yes, no, I’m not sure sort of question.
So we’ll give a few more seconds for that poll, for your responses. Great. I think you can probably close that when you’re ready, Danny, and we can see what people said. Okay. Interesting. So yeah, you can see here on the first one, you know, the plurality said there’s like no, no policy. Some actually have some first for students. On the second question, you can see that by far the majority say that there’s not approved tools for students and teachers. And the third question is whether your school district provided professional learning for AI, 50% saying yes. And then the other 50% either no or I’m not sure. So thanks for putting that in there. I think that tracks what we’re finding in our research as well.
I’m gonna hand it over now to Sarah who’s gonna dig in a little bit more to help us understand from the qualitative data some use cases.
Sarah Nielsen:
Awesome. Thank you, Drew. I’m gonna go back. Great. All right. So I’m gonna share with you all some of our interview study findings. So this is with those 15 focal teachers that Ann mentioned earlier, along with one to three of their colleagues. The questions that we were really interested exploring in these interviews were how are teachers using GenAI to build new types of learning experiences for students? And also what policies, resources, and messaging in schools and districts are shaping teacher GenAI use?
Alright, so on this slide, you are looking at a framework, and this is the framework that we use to categorize teachers’ different use cases. So in our focal teacher interviews, when teachers would tell us about a way that they use generative AI, what we did after that interview was categorize it into one of three categories, substitution, amplification, or transformation. The flow chart that you’re looking at on the slide gives you a sense of how we made that categorization. And I’m just gonna tell you a little bit about each of these categories. So the first one, substitution. So if something was categorized as substitution, that meant that it didn’t necessarily provide better learning opportunities for students. So an example of that would be if a teacher was using GenAI to write assessment questions, create a worksheet, create a PowerPoint, that would be categorized as substitutive.
The second category, an amplified use case, would be if that use did provide better learning opportunities for students, but it could have been done without the GenAI technology. So an example of this is one of the teachers told us that she used GenAI to adjust the reading levels of particular texts for her students. So that did lead to better learning opportunities, but it could have been done with a different technology. And then lastly, transformative use also led to better learning opportunities for students but it wouldn’t have been possible without GenAI. So an example of this, one teacher told us that they would use GenAI to create in the moment extension problems for students based on questions they were asking during the lesson, based on just interest that that student has. And so that’s something that wouldn’t have been possible without this particular technology.
And I just wanna name that this is the way that we decided to categorize these different use cases. There are a lot of different ways you could have categorized, but we chose to really center what was creating better learning opportunities.
Alright, so this stacked bar chart that you’re looking at here gives you a high level view of the different use cases that we learned about in the 15 focal teacher interviews. So along the X axis on the bottom, you can see the different categories of instructional tasks that we asked teachers about. So there were five categories. We had planning, instruction, assessment, professional learning, and then other administrative tasks. You can see across all of our interviews, the majority of the use cases that we learned about were in the category of planning, which isn’t surprising to us. It mirrors what Drew said about the survey results. And it’s also what a lot of the tools out there are catered towards supporting teachers with planning.
What was interesting, out of those close to 35 planning use cases, they were pretty split into substitutive, amplified, and transformative use. For all four of the other instructional task categories, the number of use cases that came up were all a little under 10. What was interesting to us about those is for instruction and professional learning, the majority of the use cases that we learned about in each of those categories were transformative in nature, meaning they led to better learning opportunities for students, but they also wouldn’t have been possible without GenAI, which is something that we’d be interested in exploring further in further research.
Okay. So we’re gonna dive into some different cases. So we’re gonna dive into the qualitative data, and we’ve pulled out some cases that we thought would be interesting and start to show the patterns that were emerging from all of our data.
So the first case is with a middle school science teacher. And this teacher told us in her intake form that she did not use AI at all. And so in her interview she told us, “Why should I invest my time and energy in the summer or on Friday afternoons? I’ve each lesson in my Google drive by subject. I go to that period. It’s easy, it’s done. I can just hit print and boom.” So we’re curious what was going on in this teacher’s context. So this teacher taught in a rural context. She was the only science teacher in her school. The nearest middle school science teacher was actually 30 miles away. And she also told us in her interview, “I have no more tech coaches. I have no more instructional people. I have no more. They’re all gone. Those are a thing of the past because of budget.” This teacher, there was, we did also interview an ELA teacher at the same school. And it was interesting because that ELA teacher was using generative AI for planning, but this middle school science teacher, she really didn’t see the point in exploring because she had all of her plans from previous school years. So that was sort of interesting to notice.
The second case is with a high school math teacher. So this teacher had a lot of use cases and most of those use cases were categorized as substitutive. And so this high school math teacher would use ChatGPT to separate concepts into four different levels of understanding for standards-based grading. He would create worksheets and PowerPoint presentations. He would create a list of questions for a YouTube video when he had a sub. So a lot of use cases. And this teacher told us that he did have professional learning communities at his school, so there was some collaboration available. But he told us, “It’s very vague on what those PLCs are. No one wants to go talk to the other people because we all have too much to do. We don’t really work together that well.” And so this teacher didn’t have a ton of collaborative partners to work with. And so it was interesting to note that there was a lot of turning to GenAI to create materials to use in the classroom.
And the last case we’re gonna pull today is with a high school science teacher. And this high school science teacher had a very interesting way of using GenAI actually with students. So what he did was he said, “We’re gonna make a lab. Let’s use AI. So AI writes the procedure based on a student’s question and then we’re going to try it, whatever it says. Students don’t get a lot of experience in actually developing and creating labs and conducting labs generally. They’re given a lab investigation, you know the result because you’ve seen it done a trillion times. It’s more fun to not know what’s going to happen, more valuable and authentic to science to actually do something that we don’t know the answer to.” So we are really curious what was going on in this teacher’s context to support this kind of use.
So we actually interviewed the director of technology innovation at the same district and he told us that, “We have a bunch of approved AI tools. Leaders come in, figure out who needs this. When do they need it? And we decide we’re gonna do PD with all staff in the fall, right when we come back to make sure they’ve all seen it. And then we always follow that up with some kind of asynchronous support piece or additional training.” We also interviewed a teacher at the same school and she told us, “We’ve had one email from our technology person about AI. It was like tools you can use, whether or not they’re safe for our district to use them, guidelines surrounding it.” She also said that, “Collaboration was a real big focus for the past five years. We had an MTSS person come in and do like this whole program about collaboration.” And so this was really interesting to us and this sort of showcases patterns that were emerging around what we noticed between how teachers were using GenAI for their instruction and what was going on.
Were leaders talking about AI in their school or district? Were there approved tools? Did they have access to collaboration and professional learning? And so I’m gonna turn it back over to Drew, who’s gonna sort of synthesize what we saw across all of our interviews.
Drew Nucci:
Thanks, Sarah. It looks like in the chat, people are already starting to pick up some patterns. Basically we looked at the use cases and we looked at how people were characterizing the social and institutional context, and some themes emerged. And the three themes that emerged were policies, resources, and messaging. And so I’m gonna just, I’m just gonna touch on the end points here. It’s a continuum. I’m gonna put a touch on the end points ’cause we’re out of time.
But when we think about the policies that we saw, the people who were not using AI at all or using it in various sort of substitutive ways, like make a worksheet, were often in environments where there were no clear tech policies or restrictive tech policies and really limited professional development and collaboration structures. Whereas the more advanced users who are really making new learning opportunities for kids, they were in environments where there were clear instructional technology policies and guidance that encouraged experimentation and innovation and ongoing collaborative professional development structures with a focus on using the tool not just for efficiency, but actually for instructional improvement.
When we think about the resources, perhaps unsurprisingly, those with limited AI use really were in places with limited resources. We even talked to people where they didn’t have internet connectivity in their whole school. So you’re talking about limited AI tool access, limited time and money for professional learning, whereas abundant resources and even student facing tools and incentivizing teachers not just to be in professional learning but lead it, characterizes the context in which we saw more advanced use.
And then in terms of messaging, this is really sort of a cultural element. In the cases where people weren’t using AI or had very substitutive use, there was real vague messaging about AI in the school and a lot of narratives about student cheating without really digging into what is the purpose of what we’re doing here? How does this maybe change math and science? How does it change how we think of what it means to get kids prepared for the world? There’s also sporadic messaging about professional learning and really a separation between teachers and instruction and leadership. Whereas in the more advanced cases, there was clear messaging about AI and really AI tied to instructional goals. Leaders were using AI, were public about their use, and were public about their experimentation. There was a value placed on teacher professional learning, innovation, and collaboration, and really a group narrative. This is what we do here, right? We collaborate, we innovate.
I just wanna put this to you now when we think about, in the chat, love to hear your ideas, policies, resources, and messaging. What should your district most attend to right now and how to promote teachers to use AI for instructional improvement? Just pop that in the chat when you’re ready. Might be harder to pick just one. All three. See, nailed it. Yeah. Also, like we should admit that these also overlap, right? The policies send a message. The policies involve resources. There’s a good reason to pay attention to any one of these three.
We found in our conversations that actually the hardest one to budge was messaging. Because messaging implies a cultural element that is really hard and requires like a sustained leadership initiative over time. Love to keep putting your ideas in the chat.
I just have one more idea before Ann closes us out with some big takeaways. And it’s that we talk to the teachers about what kind of professional learning would be most useful to them. And they said that the professional learning that would be most useful to them would be content specific about math, about science, hands-on, collaborative, and throughout a school year, focusing on using it as an instructional technology. This is unsurprising to us because it’s really consistent with all the research on high quality professional learning. We also asked some of the more advanced AI users, what kind of professional learning should we design for people who are brand new to AI? And they said, first create buy-in from those AI avoiders by showing how GenAI can help with their pain points. Then once they see what it can do, leverage that towards instructional improvement. Provide actual time for teachers to explore and experiment together, and make sure that teachers are sharing their learning.
All right, I’m gonna pass this back to Ann now, who’s gonna help us with our key takeaways.
Ann Edwards:
Thanks, Drew. So I’d like to begin our closing with a couple of takeaways from the exploratory studies that we shared today. And none of these are gonna be surprising, just sort of doubling down and reiterating them. First, and again, not surprising, but I think still worth noting, that while STEM teachers’ AI use is still quite new, it is growing very rapidly. And their use currently is primarily focused on materials creation, as in planning. But there is lots of opportunities for transformative use in lots of different ways, as we’ve articulated and seen.
Second, the interview analysis shows that the combination of structures supporting teacher collaboration, robust professional learning alongside clear AI use policies and explicit guidance on adopted tools, all taken together can support STEM teachers to use AI toward building improved learning opportunities.
And then just like Drew just said, finally, when asked, teachers clearly say what they want in PL and that is that PL is clearly grounded in and relevant to their content area and the curriculum that they are teaching. PL should be collaborative, giving them opportunities to really engage and learn from one another, and it should be sustained. They want PL opportunities that extend over the course of a year and not just one-off workshops. But that is… That’s new, especially in the GenAI space.
So if you’re wondering what this looks like, we have a possible approach to share with you. Just this past summer, we piloted a program to support teachers to learn how to use AI by teaching them how to build their own AI chat bots to improve their instruction in ways that were most meaningful to them. What we saw and what teachers said was that they learned about AI through the sandbox. They learned to build AI bots, as was the goal. And importantly then, they learned how to plan for more meaningfully engaging and relevant tasks where students talk with each other about important ideas. So if you’d like to learn more about that particular experience from the summer or how we can support your district’s or school’s AI learning or planning, we have a form here. It’s the AmplifyGAIN interest form. The link should be in the chat.
And then finally, we wanna thank you for joining us. Please reach out to us if you’d like to learn more about the studies, the work of the AmplifyGAIN Center more broadly, or about our research and services and the use of GenAI in STEM education more generally. Our email addresses are there. We also invite you to complete our feedback survey. I believe the link is in the chat. And we will go ahead and post a recording of this webinar, along with transcripts at WestEd.org soon for you to share with your colleagues.
And now I’m gonna hand it off back to Danny to close us out.
Danny Torres:
Well, thank you, Ann, Sarah, and Drew, for a great session. And thanks to all of you who joined us today. We really appreciate you being here. You can check out recordings of our past Leading Together webinars. We’ve covered a range of topics, including literacy, assessment, special education, and mathematics. And there’s more to come. We’re working on our schedule for 2026 right now, so please look out for future dates.
To access our Leading Together webinar series recordings, visit us online at WestEd.org/leading-together-2025. And finally, you can also sign up for WestEd’s email newsletter to receive updates. Subscribe online at WestEd.org/subscribe or scan the QR code displayed on the screen. You can also follow us on LinkedIn and Bluesky.
With that, thank you all very, very much. We’ll see you next time.