aiEDU Studios
aiEDU Studios is a podcast from the team at The AI Education Project.
Each week, a new guest joins us for a deep-dive discussion about the ever-changing world of AI, technology, K-12 education, and other topics that will impact the next generation of the American workforce and social fabric.
Learn more about aiEDU at https://www.aiEDU.org
aiEDU Studios
Zach Kennelly: My civics students used AI to create a voting app
When civics teacher Zach Kennelly first encountered ChatGPT and DALL-E, he immediately recognized their transformative potential for education. As one of the first AI Trailblazers in aiEDU's fellowship program, Zach has reimagined what's possible in the classroom by positioning AI not as a replacement for human thinking, but as a collaborative tool that empowers students to tackle challenges they care about.
Throughout our conversation, Zach shared how his diverse background in political science, sociology, math, and science provided the perfect foundation for integrating AI into his teaching at the Denver School of Science and Technology Public Schools (DSST) network. Despite initial roadblocks after the school blocked AI tools due to privacy concerns, Zach persisted in his belief that providing students with AI literacy was fundamental – students without AI literacy would soon be competing against peers who were becoming fluent in these technologies.
The results speak for themselves: Zach's civics students created VoteWise Colorado, a voter engagement app that caught the attention of the Colorado Secretary of State. Rather than traditional assignments where teachers dictate knowledge for students to absorb, this project centered students as experts in their own community's needs. And the lessons learned went beyond academics; one student confidently declared "I could run a tech company," while another reflected how "I care a lot more about things than I knew."
Are you ready to explore how AI might transform your classroom? Zach recommends starting in areas where you have expertise, focusing on low-stakes experimentation and remembering that the goal isn't to replace traditional skills but to elevate them. As we navigate this pivotal moment in education, teachers who thoughtfully adopt these tools aren't just preparing students for an AI-driven future – they're addressing fundamental questions about what it means to learn in the 21st century.
Learn more about Zach Kennelly and DSST Public Schools:
aiEDU: The AI Education Project
so we're here at another early recording of ai edu studios, with someone that I've been really looking forward to interviewing uh, zach cannelli, one of the first ai trailblazers, which is a cohort of innovative teacher leaders that we brought in as part of a six-month fellowship, and Zach was someone who you know. Very early on, we realized, you know, this is exactly what we, when we closed our eyes and envisioned like, what could the impact of creating this sort of experience for teachers be? It's not just, you know, training and impact, but really like creating champions who can drive this work for themselves, and so I want to dive into that. But, Zach, can you so welcome, Can you give us just like a?
Alex Kotran (aiEDU):You know it doesn't have to be your whole story, but you know how did you get into education and then you can just tell us about you know what you teach, where you teach, what your school is like, to sort of help us paint a picture of what your day to day is. I know that you actually have a career shift within your system that you can talk about as well, but maybe just sort of like sharing up until that point before you get to the sort of like what you're going to be doing next.
Zach Kennelly:Yeah, thanks for asking First. So excited to be here. Really appreciate you asking me to be on, alex. Yeah, what a journey.
Zach Kennelly:So I do think a big part of this story is that my undergrad was in political science and sociology at CU Boulder, and then I went through TFA and did math and science. I got my master's in math and science, and so I think that broad training was really important to my passion for AI because it allowed me to like apply different understandings of content and pedagogy with dexterity, and so that was really important in that I had taught in middle school and high school and math and science and social studies and civics, and so really early on in the generative AI boom, I saw it and knew that this really, really mattered. I had always been tech forward, but I hadn't been incredibly focused on AI, and so that process of seeing this really matters for society, for young people and for our system, and especially what this means in a civic context, really ignited my imagination. And then I just started to play right, getting connected with different communities. Aiedu was a key community.
Zach Kennelly:The Trailblazers cohort really allowed for me to connect with people across the country because originally it was blocked. Almost all AI use was blocked at DSST and I really understand where that came from. It came from a purpose-driven place, and so the opportunity to push that was really powerful. So I just mentioned DSST a little bit about that. Denver School of Science and Tech we're a charter network in Denver and focused on urban and suburban, serving predominantly low-income students of color and STEM career fields had some really great data come out recently that our graduates in the STEM field are earning significantly more than peers in other fields, which is really exciting, and so we see building AI competencies as a very important part of ensuring that the young people of Denver and beyond are future ready.
Alex Kotran (aiEDU):And so you teach social studies. That's right, my heritage psychology and senior civics and senior civics, not social studies broadly, but actually civics, which is really important. And when you say, like this generative AI moment, what was it like? What was your first interaction with generative AI? Was it ChatGPT? It was.
Zach Kennelly:It was ChatGPT and Dali chat gpt, like when it first came out, it was chat gpt and dolly. So I was on a fishing trip with my father and jamming out to some economist uh articles you know audio version, and they talked about this emergent capability. I had, you know, been listening to a bit of what's happening in generative AI. This could matter, but as soon as the emergent capabilities came out, that's where I was excited and curious enough to go spend several hours trying to figure this out and play. And so right away I built an image on Dolly.
Zach Kennelly:I remember running upstairs it was like this beautiful hawk over Longs Peak, which is a really famous peak here in the Colorado Front Range and running upstairs and showing it to my wife and being like, oh my gosh, this is incredible. The hawk had three heads. So it was also very interesting in where it was precise and accurate and where it was really wrong and where it was messing up. Then also started to play with chat GPT and building out different use cases in education in my life I love to write, but I also kind of struggle with syntax and spelling. I've always struggled with that since I was a little kid, but I also love to write. I scored in like 98th percentile on analytical writing in the GRE, but they don't score spelling and they don't score grammar, whereas if they did, I would have scored much lower, and that was a huge moment in my life because I had always struggled as a writer. I got like my best grade was a C in high school and so yeah, that is a really important part of my life.
Alex Kotran (aiEDU):You don't expect me as someone I kind of I would have assumed you were, like the straight A student. That's right, yeah.
Zach Kennelly:So no, I was a. I was a football blockhead. I only did enough school so that they would let me put on a helmet and go smash my head, and so actually that's an important part of my story is like did that and then went back to college later and really embraced my passion for writing and analytical writing, especially in political science, and so what I really loved instantly about interacting with chat GPT was being able to take the incredible number of thoughts in my brain and like get them out quickly and help them refine quickly using AI. I was never super excited about like AI doing things for me.
Alex Kotran (aiEDU):It was not ever really a big efficiency play no-transcript part of understanding sort of how we got to this zeitgeist moment. It's like it's jarring when you see this technology thing create language. You know, when you ask it to sort of like, write an essay and it produces an essay and you're just like it, it. It manifested in a really sort of visceral way something that was previously really arcane. Right, like you know, machine learning was around, like it was touching our lives. It was recommending the content that we consume, um, you know google maps, machine vision, you know it was in our phones, in our computers, but it was mostly invisible, and I think also the chat interface. It made it something because it's like interactive and you could almost have a conversation. You can have a conversation with it.
Alex Kotran (aiEDU):It opened the door to people that weren't technical and, as someone who also, you know, like I, was a history nerd, in high school, social studies, ap European history was my favorite subject. I love reading history and so you know I'm a humanities kid and you know language models because they are, you know, working with language. It was a. It created for me a very accessible way to interact with. You know this technology, um, so so you joined, you, you joined. How did you hear about the AI trailblazers. Like did you just apply?
Zach Kennelly:great question. First, it's so great to hear about your humanities background, right? I? I kind of thought maybe the CS background, um, so I really I I love seeing, uh, the way this has grown for you, right? I'd love to also at some point hear how you saw the need for AIEDU. That said your question. How did I hear about AIEDU? I actually got on a podcast with. It was with Christian Pinedo, patty Quijones of Colorado Education Initiative and Adil Khan and they invited me to. Hey, I heard Zach's doing a lot with this Hop on the webinar and then saw Christian Pinedo and then I can't remember exactly how that came about, but got referred to AIEDU and jumped in and applied and so grateful that that relationship has blossomed and we've had the opportunity to work together.
Alex Kotran (aiEDU):Yeah, excellent. So what was it like? I mean, you had you had you had an opportunity to sort of like spend time with other educators prior to that, you know, in sort of like a space where everybody's sort of like working on AI together. Is that like an opportunity that you had in DSST?
Zach Kennelly:Not at all.
Alex Kotran (aiEDU):What does DSST stand for?
Zach Kennelly:Yeah, denver School of Science and Tech. We also now have Aurora Science and Tech, a neighboring city, and so, and this is a charter.
Alex Kotran (aiEDU):Is it a charter system or a charter school? Within a charter system, that's part of the public school district.
Zach Kennelly:Yeah, you nailed it. We are a charter network within a public school system, so DPS is the umbrella. We operate in DPS, we have DPS facilities, dps transportation and then we're a charter network within. There are also several other charter networks. So Denver has Choice, which is a really important part of our story is the ability for young people to choose where they go, and we are a public charter, so totally publicly funded. Anyone can get in.
Zach Kennelly:We do have wait lists and our STEM mission has been a really important part of that process and so, really on, even though I teach civics and AP psychology right away, I saw a natural relationship, of course, with the humanities, but also with STEM and AI. So I didn't have the opportunity to talk with a lot of other teachers because actually it was blocked. At DSST there were significant privacy concerns. There was a lot of questions about like how do students sign in? How is that data being used? What are the risks, and I think all of that was right.
Zach Kennelly:In the initial Trailblazers cohort, not only did we explore, you know, ways to help young people understand this content, but also thought partnership and like how, what are the risks?
Zach Kennelly:How are other districts approaching this?
Zach Kennelly:What are the ways that we can talk with our district and help them understand the way that I'm seeing this, and the way that I'm seeing this is very much an equity issue.
Zach Kennelly:We have to get young people the ability to use this not necessarily, you know, require them to use it, but build the skills to use it so that they can understand what they're working with and the idea that these are very much going to be the challenges the challenges of the age of AI are going to be the challenges of the young people that were in front of me right then and I was teaching all seniors and I was looking at them, I was playing with AI, I was understanding the potential and I knew so many of them were going off to college next year and were going to be competing against other people who were learning about this and we're learning how to leverage it and we're learning how to use it and we're learning about the pitfalls and the opportunities, and so I felt really compelled to get after it, in collaboration with this group, and help DSST like really orient to the power and peril of leveraging AI, so that we can put young people in a position of power.
Alex Kotran (aiEDU):Yeah, and so you know, this is something that we had spent a lot of time talking about, and it was abstract, right, Like the idea of like, okay, what does it look like to actually give students or create learning experiences for students that allow them to both engage with the technology while also building skills, the durable skills that we know are going to be critical? I guess it's a pun, because critical thinking is one of them. Yeah, what does that give? Can you give an example of how you brought that to life, especially as a civics teacher? Right, so you, you know. I think most people assume that you know the place where you teach students about AI is a technology class or a computer science class, and I don't know that they would necessarily think of civics as a place where you do that. And yet you figured out a. Really you know like more than one, but you know one really powerful student project that ended up getting national attention. Yeah, can you share more about, sort of like, how you brought this to your kids?
Zach Kennelly:Absolutely. So really early on. Okay, so I've always had sort of this problem with the way education approached both the internet and social media. Uh, I went to Columbine high school down here in in Littleton Colorado. Um right, challenging and incredible experience. So many people, uh, filled with deep care, incredible educators at the same time coming of age with the internet. So many people were like, don't use it same thing with social media, we're not exploring it, just stay off of it. Right, it's not, it's, it's not going to be helpful. I really wish that I had had more educators in my life and in all of our lives.
Zach Kennelly:Invite us into a conversation, right, hey, this internet thing, this social media thing, might be a big deal. Let's think about the implications. And so, really early on, I felt compelled to be that for young people and I think AIEDU Trailblazers especially helped me find a lot of that language and approach. And being a civics teacher, you know I've taught civics and ap psychology goes really interestingly hand in hand. So I've taught a lot about the internet, social media, how social media is so good at our the reward system in our brain and so good at nudging us to more extreme content, because that's what's highly engaging, and that's how social media companies tend to make money, and that there's some great things about that, but there's also some real perils about that. And so, right away with generative AI, we saw it as an opportunity to invite our young people into critical conversations about the implications for this technology, way beyond the tech sector and in their lives, focused on power and empowerment, and what I mean by that is a gap that we think is in CS education and engineering. Education in general is like a deeper focus on the human impact of these technologies, and so we saw it as an opportunity, in this STEM community, to help our young people really think about, you know, what are the implications for the future of civic society and the future of humanity when we're thinking about this new technology that can generate, right now, language, but it may be able to generate a lot more as the technology advances.
Zach Kennelly:And so I wanted to position young people as experts in their lives, experts in the problems of the community, and then be able to work on those. And I think what's really important about this you talked about our big project is that we authentically centered young people. What I mean by that is we started out with okay, what is power? Then we helped them look at the technology, and this is where we used a lot of AIEDU materials. One of the best examples is the 29 AIs of Washington DC no-transcript plays out over time.
Zach Kennelly:And then we looked at you know, what are the implications for human relationships? We've worked with the Rhythm Project for that and then we had students build bots on PlayLab and these bots were actually focused on their story, their expertise, and that came back to when we were looking at power. And then we put students into groups. We said okay, a bunch of you built bots on voter engagement. Do you want to build a voter engagement app for Colorado? And students felt really excited about this. The thing that we're seeing here is that so often we're telling students what to focus on problems that matter to them, and that's what happened with Vote Wise, colorado. Young people were focused on building a voter empowerment, voter engagement app. We put them in collaborative teams and they blew us away. We thought it would maybe be a failure and it wasn't. They built something incredible.
Alex Kotran (aiEDU):Yeah, I was just talking to this venture capitalist who he had this really interesting idea, which is, you know, the key to preparing students for the age of AI is all about equipping students to basically be their own CIO, and you know, what a chief information officer does in an organization is like they're making decisions about what technology the organization uses. And so his point was we need to give students the ability and the skills to evaluate different tools and um identify which tools are the right fit for the goals that they have. Um and it sounds like you're in PlayLab for our audience is a platform. It's a nonprofit organization that allows teachers and students to create. If you've built like a custom GPT, it's sort of like that, but it's, you know, like a model agnostic. So was this an idea that students had that you brought to students? Is this something that the students had sort of? Was it their idea that they came to you with? How did you get from having the students just explore to this very clear idea of what you wanted to help your students build?
Zach Kennelly:It's a great question. Okay, so what we used to do? We always used to do something called story of self us and now. So our big thesis for a long time was that the greatest way to empower students was to teach them how to tell their story, or a story in a way that impacts others, that can drive another to action. So we use Marshall Gans story of self us and now and really like drive that.
Zach Kennelly:We thought that AI teaching students how to leverage AI was a better way to help them tell powerful stories, and so we knew we wanted to do a project instead of our story of self, us and now project, which we had done for years. We wanted to do an AI, human-centered AI leverage project and students chose voter engagement. So we had tons of things that students proposed and built bots around. Students were really interested in empowering immigrants with information about how to navigate our extremely complex immigration system. Students were really interested in financial literacy empowering folks with financial literacy bots. Housing is incredibly important. Young people wanted to help people with housing and identifying resources for affordable housing, but the most popular was voter empowerment, and so many kids were talking about.
Zach Kennelly:I have family members who are eligible to vote, who want to vote, but the process feels daunting, and so they started to build bots around it and we formed a hypothesis that we could build an app that was incredibly powerful that could drive results. We did have a secret ingredient, essentially, which is something called Boxcar. We had a local AI expert reached out to us wanting to work with us, and he is truly mission aligned founder of Tin man Kinetics here in Denver, and why this matters is we used PlayLab. We used Claude to build out our prototype of the app Students, genuinely, you know a room full of kids just building out the app in roles. And then Boxcar was used to code the back end, which is a big challenge right now, which is why we were able to launch a enterprise grade web app from a high school classroom. It's incredible, wow.
Alex Kotran (aiEDU):It was truly student driven. So what's your vision for you know what comes next. I mean, you have one amazing you know app that or a project that students have really leaned into. But I think you have you talked with me, you know, when we were at the ACO GSB Summit together earlier this week. You have a lot of other ideas in terms of like how to continue, sort of like leaning into this way. Yeah, what's your vision?
Zach Kennelly:Yeah, it's a great question, OK, so two we just wrapped up our second round of this and students built. We were focused on empathy fatigue. So the problem that we started to focus on is, because of social media, so many people struggle to empathize really with folks, and we have a large population of young people who come from immigrant backgrounds, have immigrant family members, and so many of these young people are just like, oh my gosh, my family is like suffering, my parents you know, they're incredible people. They've come here, they've gone through this, a lot of people with housing insecurity, first gen college student. And so what they did is they worked in teams to create artifacts mostly on Claude that tell stories, to create artifacts, mostly on Claude that tell stories. So the idea is it's easy to judge someone. It's really hard to judge them if you see through their eyes the choices they have to make every day, and that's associated with sensory details to trigger empathy in the brain, and so that's what students did and that one didn't get as much press. It's not an enterprise grade app, but there's a lot to be taken from it and students really love telling their story, and so what we're really starting to think is how do we put this together in a way that allows people to do this at scale, because what we've done right is it's challenging to replicate.
Zach Kennelly:You know, I've been doing this for a long time. I have a broad set of experiences. I had an incredible teaching partner in Gianna Giraffo, who's just like was in software sales before being a teacher, left, you know, a much more lucrative career to come do this, and so a lot of things came together for us to do this. But we do believe that we can sort of lay a foundation and collaborate with incredible organizations like AIEDU to help other people do things like this and ensure that people are seeing that it's not about efficiency, it's not about doing you know things faster or an old model faster, but it's about empowering young people so that they have the motivation, the desire, the imagination and creativity and agency to solve problems in their world and drive impact. And when that happens, we see that it solves all these problems around, like it really addresses attendance problems, interest problems, and so students can do great things and they get excited about it.
Alex Kotran (aiEDU):Do you? It's interesting that you mentioned solving attendance problems. That's not intuitive, but, you know, absenteeism is a really big challenge in districts across the country I think there was one of the big districts you were talking to is like 40% of the kids are not showing up to school, and this is like a multidimensional challenge. There's lots of things that feed into that, but one of the consistent things that we that you see from surveys and the research is that kids just don't feel like education is relevant. They just don't understand and maybe even rightfully, are seeing that, you know, the stuff that they are being forced to learn is just not connected with their day to day experience. Experience. But, yeah, what's behind the? This phenomena of students, like you know, AI, actually sort of like drawing students into education? Is it just that connection to sort of like what they are actually experiencing? As you know, digital natives?
Zach Kennelly:This is such a great question, alex, because, okay, so I'm observing now one of my AP psychology. One of them is first period and first period, like you know, can be really challenging to get to all of those things, but my attendance in my civics classes is stronger than AP psychology and, like AP, psychology is incredibly interesting. You know, we have great attendance in those classes overall. But even at that scale right, this AP course versus this highly creative course and we do some of that in AP, but it's an AP curriculum, right, I'm really focused on the curriculum, and so what I think we're seeing is when we position young people to work on things that they choose as relevant, while also building the critical thinking agency, creativity, imagination and technical skills to create it solves the challenge, just like you said, of relevance. We've known, right, for I don't know a really long time, that relevance matters. I was just teaching Piaget and you know, context In context, learning matters, and so, when I think about this, I think the biggest thing that we've been doing, and with good intention, is lying to young people and telling them that the process to success is doing what you're told.
Zach Kennelly:Do what you're told. I'm going to tell you how well, you did it. I'm going to give you a grade and then if you do what I told you to do, you're successful. And kids are saying I don't believe you and they're starting to see like actual, real success is our ability to see challenges in the world and address them relentlessly, iteratively, and collaborating with people and AI. And so where I see AI coming into this and I've heard some people who are critical about like the lower friction I don't want to lower friction of learning, but I do want to lower the friction to collaborate Like to be a great teacher in a problem-based setting is incredibly time intensive. Right To set all the things up, to do all the grading, to give all the feedback On our projects. We gave zero grades, not one. Students graded each other and it turns out they grade each other harder than we grade and the authenticity is deep and rich.
Zach Kennelly:And so what I think is the thesis here is we position young people to work on things that matter, we equip them to collaborate between each other and on with AI, and then they are accountable to one another. But, most importantly, that authentic audience right, it got really real for them when I'm like, hey, the secretary of state just called and they've got some feedback. Those types of moments are so powerful. Hey, the CEO just reached out of DSST Public Schools, nella Garcia-Urban. She used our app to vote. She'd love to celebrate y'all and tell you about her experience.
Zach Kennelly:Those things fueled our young people in a way that I haven't seen in my 14 years of education. We've done a lot of really great things, and so that's one of the things that keeps me on fire about this work is the opportunity to position our young people to be creators, to work on things that matter the most to them, and to get away from so much of the narrative of like what's wrong with the young people to let's create a system that empowers them to do things that they love and care about yeah, this was um, like victor lee at stanford did some research and one of the things that really caught people's attention is they found that this assumption that chat gbt is causing students to cheat more is is flawed.
Alex Kotran (aiEDU):It's actually that you know students are using chat GPT to work on their homework and you know, at least for me, I think it's important not to write off, like when teachers are stating concerns about that they're like we should trust their instincts and the answer isn't just like oh, you know, you just have to allow students to use chat gpt and be okay with it. Um, you know, I think there's that's a conversation I want to have with you about. Like, what is, what does guidance look like to educators who are trying to think about that? Um, but victor's research really suggests that the rate of cheating is about is still the same as it was before. In fact, historically it really hasn't changed that much as different technologies come in and even, like I think with Chegg, they had I forget exactly what it's called, but you can go to the back of you can get answers to a lot of, like the homework assignments. Do you know what I'm talking about? This is like oh, I know exactly what you're talking about.
Zach Kennelly:That's exactly right.
Alex Kotran (aiEDU):But ultimately, the root cause of cheating is that students don't feel like what they're learning is relevant um, they feel like they're just being given busy work and um, and so that's. That's encouraging, insofar as it shows a path to how we address this. It's it's not that we need to entirely ban AI or, you know, completely change the way we teach, but it does mean that there's an opportunity to evolve, maybe how we teach, and you just described maybe one really cogent and sort of vivid example of what that could look like. But I think the cool thing about this technology is that any teacher, no matter their subject, um, as long as they know how to use the tools, they will be able to. They can both design, you know, assessments and and projects that, uh, the the tools by themselves can't just solve on their own, um, but in fact, teachers can probably use the tools to help them in that endeavor. Um, which is sort of like the meta aspect of the technology is, like you know, I don't believe that ai is going to solve the problems that it's going to create. I'm not a techno optimist to that extent, um, but maybe I'll put it a different way and I'm curious for your take on this, I I feel like the.
Alex Kotran (aiEDU):The only solution to cheating is for teachers like, like right now, the the, the fact that students are able to use AI to get it like around their teacher's assignments. It's coming from the fact that they are just more experienced in the tools than their teachers, and so the teachers are at a disadvantage, because it's very hard to design like a, an llm resistant homework assignment if you don't really understand what it's capable of. And I see this with, like you know, teachers who say, oh, there's, like you know, I put a trojan horse in my assignment. This is, like you know, the idea that you, if you have like a writing prompt, you in, like, you add like white text and like white font, you know something like, oh, add the word Frankenstein somewhere in the in the output, and that probably works for a period. But then the kids figure it out and they are utilizing, you know, social media and communication tools that will, you know, allow them to quickly get ahead of that, and so these are just band-aids. The only thing that a teacher can do to really stay ahead is to make sure that they themselves are, you know, proficient, if not power users. I don't know that they have to be power users, but they have to know what the tools can and can't do.
Alex Kotran (aiEDU):Another example is when I hear educators say, oh well, I just have students. Uh, you know, I say it's okay to use chat gbt and then I ask them to critique the outputs. Um, so, like you know, like have the ai, create an output and then, like you know, sort of like, go through and identify, you know, the mistakes or, you know, provide some commentary or another example of self-reflection, and to me that just demonstrates that teachers don't realize or haven't really thought that like thought a few steps down the the road where it's like you can just ask chat jpg to critique itself, um, you know, that's literally just one extra prompt that you like, sure, like you've added a step, um, but by itself, you know, and and a self-reflection is the same thing.
Zach Kennelly:You can.
Alex Kotran (aiEDU):It's not that hard. It maybe takes a few extra sentences in the prompt engineering. It's not maybe a single sentence prompt, but you can probably get a pretty damn good reflection, and and maybe you and I disagree about this I'll make a statement that you can react to. So I really believe that you talked about friction. I'll reframe it as like productive struggle. I think what teachers are getting right is that AI does make it easier for students to shortcut work, the result of which can mean that they're just not going through the productive struggle that we know is so critical to their development as learners.
Alex Kotran (aiEDU):And so like for me, as someone who did a lot of writing in college and high school, you know sitting in front of a blank piece of paper and struggling with, like, how do I start this essay? How do I get what's in my head onto that piece of paper? It is the hardest part of writing, but it's also like the. Maybe the most important part is like building the. You know persistence to, to get past the writer's block, and if your instinct is always just to like, oh, I'll just open up chat GBT and have it create an outline, and then I'll like, go through the outline and provide some feedback and then boom, I have like in a first draft and I'm editing the draft and that's. And actually I was just so the the same vc that I mentioned earlier.
Alex Kotran (aiEDU):Like he, his model is he has two different lms create outlines. He shares each outline with one another and then he has them sort of like, incorporate them and create the better you know, create like a revised outline, and then he gets the draft. Um, but even that I, you know he and I were reflecting that part of why he's effective at prompt engineering and using AI to help him write is because he was already a really good writer. And I worry about students that you know, before we get them to use AI to augment their abilities as writers, they need to kind of build that core writing expertise. And I think it's different than prompt engineering. Right, you know writing is necessary to be able to prompt, but it doesn't necessarily go back and forth. I don't know that if you just learn prompt engineering, that you're now also learning, sort of like, all the core skills in writing.
Zach Kennelly:I completely agree.
Zach Kennelly:Okay, that statement Okay I was hoping you would disagree, but I but I have. I have a lot of nuance here. So the first thing I have to say so I'm very lucky in that my mom was an early an ECE educator for many years, director, right. My sister is a K through five interventionist, but she was a kindergarten teacher for many years. Three kids I've got my son, like he is learning to read with paper books, and so I do really believe, exactly like you said, the daunting experience of a blank page and having to create that matters, um, the big challenge that I have. So the first thing I would say is, like reading and writing and math are as relevant as ever and possibly even more important. I want to say, okay, so I think Socratic circles, right, that required deep analysis. Those types of things are as important as ever. Types of things are as important as ever. I also think, exactly like you right, that expert writers are the ones who are able to have the taste needed to differentiate with AI, and so, at the same time, we have a huge problem in education and that's that people don't get feedback. And so, in addition, if I were to like really put all of my theses together into, it'd be like essentially two, maybe three buckets. The first is like impact focused, human centered AI leveraged creation. For young people, that's especially in subjects like history, stem, science, right. These kids should be creating more.
Zach Kennelly:Reading and writing are essential, and the best use that I have seen is immediate feedback. Now I do art. I can see people saying, well, the immediate feedback can slow things down. I've used class companion. I've seen folks working on Quill, really curious about Quill. Both of those like I can definitely vouch for class companion. The reason I vouch for it and the analogy I love to use is like kids who shoot free throws and then they find out two weeks later whether or not it went through, and I have a really large number of MLL learners who really need that feedback right away.
Zach Kennelly:So two places that I use AI for writing is I help build sentence frames and exemplars Like this is something I'm an expert in, but it takes an incredible amount of time. So sentence frames are essentially like the mortar language the language that holds together the content expertise students have. A lot of MLLs need support with this, and so we want to use it as a scaffold. They use it and then they move on from it. I use AI to build those. That's a huge use case.
Zach Kennelly:The next is immediate feedback, and so that's where young people get feedback right away on their writing and impact. It impacts their ability to improve faster, and that's an extension of me. Like, I don't want the AI to own that. I want to be able to create the conditions for what type of feedback would I give, or would hopefully somebody better than me give, and then help that young person improve faster? And so there are a bunch of places where, like kids using AI to write especially really from K to maybe nine or until they develop some place of proficiency, will undermine it long term. And so, yeah, I really agree with that and I have pushes around like how do we support young people leveraging AI in a way that was never possible? How do we give feedback faster in a way that was never possible?
Alex Kotran (aiEDU):Yeah, one of the. So you hit on something that I that opened up my mind about how to think about this, and it was like if you think about other technologies and how they impacted the way we learn.
Alex Kotran (aiEDU):People talk about the calculator, which maybe it's an overused analogy, but it's also, I think, quite apt, because I don't think anybody would say that the calculator you know, prevented us from teaching math, but it certainly changed the focus from being just, you know, going through like the raw arithmetic, to problem solving and showing your work, and so that was like a. Really that's a narrow tool and so maybe it has a narrow solution. That's a narrow tool and so maybe it has a narrow solution. But I was actually doing some. I used the deep research tool on chat, gpt and Gemini's research tool as well, and the question I had was is there any evidence that, or like what happened to the volume of writing before and after? After uh, students had access to, like you know, uh, computers and keyboards, um, and yeah, it turns out like, if you go to like the 1980s and then compare it to like the 2000s, like the 2010s, it's like uh, long form writing the, the, the average length increased by three to five x. So before it was like about, I think, per week. So, like I think before it was about like a half a page per week of of writing, um, and then, after the computer, you're getting to, you know, uh, multiple pages per week, um, so the volume increased, but that's because it was, you know, in the same amount of time, you're able to write more um and so what
Alex Kotran (aiEDU):what you're able to write more, and so what you're describing is almost like thinking about what the next iteration of that will be, where you know the homework assignment might not just be to write a self-reflection or to, you know, to write an essay about the importance of voting. You know to write an essay about the the importance of voting, um, but rather what you describe right is like a project where you're building an actual app, uh, to help people register to vote. And you know, five years ago, if you described that project to a teacher, I mean how would you do it? It would be like.
Alex Kotran (aiEDU):I mean, it's like, maybe like you know one out of a hundred kids would would have the resources to be able to figure that out. Have to know yeah, that's exactly, if you're lucky, right? Yeah, you'd have to know how to code. You'd have to have, you know, the tools to be able to, like, deploy the application. So, but it does mean raising the bar Big time.
Zach Kennelly:Yeah, I really love what you just said. Right, I've got some things to share. A couple of my favorite quotes Young woman like been through so much right Incredible level of poverty, deep trauma, straight. A student also has her own business. So like, as as incredible a human as exists. Her statement at the end of this so simple I care a lot more about things than I knew. And her statement from that came from like analyzing these really charged ballot measure statements in Colorado, informing opinions around them Came from taking ownership of how people vote. Another one incredible young woman I could run a tech company. I could run a tech company. I could found a tech company.
Zach Kennelly:And you're like, yes, that I enjoyed working with other people to do something that where nobody told me how to do it. We started and and we said, hey, we actually don't know how to do it. We started and we said, hey, we actually don't know how to do this. Like, we're really good teachers, but I never done anything in software development ever. We want to go on a journey with you. Let's get in on this. Like you're invested in the voter education. We're in voter empowerment. We're invested in voter empowerment. Let's go make this happen. And that was incredibly invigorating for kids, because so often we say, hey, here's all my knowledge as a teacher, let me just deposit it in you, and we act like that's not true, but like it's a huge tenet of modern education. And so meeting young people there as peers and saying, hey, I know that you have a bunch of expertise here and we could build something special, like let's do this together. I think there's a lot of opportunity there and I think it's going to be really hard for a lot of educators across the world and across the country to say I am okay, not being the keeper of the knowledge, and we are here together to do X, y, z, um. And I think it's really special and I feel really excited about the opportunity to do that.
Zach Kennelly:And I don't think it's all or nothing, right, I think it's. It's you know, we we've got. I love the analogy of the bridge right. When we, uh, build a new bridge, we keep the old bridge up while we build the new bridge and then we transition. And I see us very much at this moment, right now. Right, we're trying to span this gap. We know the future is going to require new skills. We've got this giant system and so we've got to be able to do both at the same time. You know, build reading, build writing. Both at the same time. You know, build reading, build writing support. Help kids build the currency they need for things like an SAT score, while also inviting them to be creators of the future they want.
Alex Kotran (aiEDU):The bridge analogy I love. I love my analogies.
Alex Kotran (aiEDU):The bridge is cool because it's like, if you know you're going to need a new bridge, you need to start building it, and that doesn't mean that you knock down the old bridge right away.
Alex Kotran (aiEDU):And so I think, for folks who are trying to make sense of their instinct about well, it's not that, you know, I push back on this whole idea of like we need to revolutionize education, because I think there's, you know, the Silicon Valley mindset of like, move fast and break things doesn't work in education. But this is one of those places where we need to move fast. I don't know if we need to break things, but we need to move fast and like, start, you know, start building and learning, um, because at some point you're going to need that new bridge and you don't want to be, and it takes time. You know bridges, which is also why I like this analogy. It's like it's not something you can do overnight, um, and so, even if you don't necessarily have all the things that you need you may not have all the concrete like, you could still start building the scaffolding and taking this a little bit far. But yeah, I think it's a really good analogy.
Zach Kennelly:I think it goes really far and I think helping educators know like you don't have to shift everything tomorrow, but if you want to, there's a huge opportunity there. Or if it's just introducing increased feedback or if it's just making things more relevant for kids, right? I, you know, on PlayLab, built a bot for AP, psychology, AAQ questions, article analysis questions. It's a brand new curriculum. I love the curriculum shift, but there was this huge obstacle I was facing as a teacher. In one drive to work I used chat, GPT, voice to generate a prompt. I then got to work and put that prompt into a PlayLab bot and I have continued to generate these highly relevant articles that I give to students right away, and the reason that matters is it's directly related to what we're learning then. So students are more interested and I think educators can start wherever they are. But we all have to start, and that's one of the things that I really appreciate about AIEDU is giving people the spot to start.
Alex Kotran (aiEDU):So what does it look like for someone who's listening, whether they're a teacher or a parent? How do they get started? I mean, I think it's like sometimes just getting the wheels turning is the hardest part, like, what advice do you have as someone who I don't know if you consider yourself a super user, but certainly a power user of generative AI?
Zach Kennelly:It's a great question, Alex, and I would love to hear your perspective here. The one thing this is actually the title of Eric Liu's book. He's a big part of it. We teach students about power, but it's you're more powerful than you think, and that is, I think, helping people see like you are more powerful than you think. Leverage AI to do what you want to do. When people see it I think it's this like daunting thing that they have to go learn. It's scarier versus if they see it as this thing that can accelerate the things that they care about, that can help them be, you know, a stronger, more powerful version of themselves, then it's really wonderful. Right? It's not about using what the AI generates. It's about taking command and making the AI generate what you want, and so helping people really see that they are powerful and that it is learnable. If we compare those two things, people will be able to do special things.
Alex Kotran (aiEDU):Yeah, I'm trying to think of how to get even more practical and I think it's like Good, good for you. I think it's like I watching this, this expert on language models, um, and he was describing what cyber security is like in the day and like in this, in this sort of age of of language models and generative ai and cyber security is basically people just sort of like playing around with these tools and sort of testing them out and seeing different like what types of prompts will like get you know what happens when you try different types of prompts and um and and in abstracted, the advice was basically the. The key to to using any generative ai, but specifically language models, is um. You have to be using it in some place where you have expertise, because you need to be able to know, because you know it's they're, they're. They produce amazing things, but a lot of times they produce stuff that's wrong. They make stuff up and, more importantly, they are really good at sounding confident, regardless of how correct or incorrect they are. Um.
Alex Kotran (aiEDU):So it needs to be in an area and it doesn't matter whether it's you know, basket weaving or history or sports or art or literature I mean using it in a in a domain where you have some expertise is key and then the other piece is experimentation um I don't know if there's anything else.
Alex Kotran (aiEDU):I don't know there's like prompts to memorize. I don't know if there's anything else, I don't know if there's like prompts to memorize. You know, at the end of the day, like you know the best the way to learn prompt engineering is just ask the language model. Hey, this is what I want to do, like how can I prompt you to get the best output? And it'll literally tell you like step by step, here's the things that it'll say, things like give me some context. And you know, um, one thing that works with some of like the older models, not the pre-reasoning models, is, you know, have the llm sort of take a persona. So if you're trying to write, let's say, like a history paper, you say, oh, you're a expert in, you know us history or uS government. That's less necessary now with reasoning models, because they actually do some of the step-by-step work for you. In fact, when I was doing my research about you know the volume of writing you know I asked for you know, this is what I want, and it actually sent a bunch of questions. It was like do you want me to focus specifically on K-12? Do you want to look at the us or are you open to sort of international uh, you know research in other countries? Um, like, what's the time period that you're looking for? And, um, you know, I think in the past you'd have to sort of like add all that context yourself. Yep, so, so it's less about learning all those specific tips and tricks, but but rather just sort of like trying things out, and what you'll find is often, you know, honestly, 75 of the time it's faster to do it, to just do it yourself. You know, it's not that you need to change the way you work and incorporate ai in every single piece.
Alex Kotran (aiEDU):I think this is the case for educators. It's, uh, it's, it might not be faster to write a lesson plan with ChatGBT, because you might write what, you might have a create one. You look at it it's like, oh, this isn't very good, and you spend some time giving it feedback and by the time you go back and forth and back and forth to get something that's good enough to your standards, you probably could have just written the lesson plan yourself. But in the process you start to figure out ways where it can be helpful, and so maybe it's not writing a lesson plan, but you might say can you and I think you mentioned brainstorming. It's like enlisting as a brainstorming partner. Hey, I'm trying to write a lesson plan about you know connecting these two topics. Let's say, you know, let you know to give your example. Uh, you know voting and civics and I want it to be something that students are. You know creating some kind of a project. Um, you know, you have your model and format for what you want to do but it. But you just say, give me like 10 ideas and you go through and maybe all 10 are bad ideas. But the process of like doing that, you might think of a good idea yourself where you might say, hey, let's double click on this one, like can you revise that a little bit? And in many cases it's still not going to be faster. I mean, this is the thing is like don't, don't expect it. It's not going to necessarily be more efficient or faster right away, but you will start to find some places. You know like for us it was in some places. You know like for us it was yep, we don't.
Alex Kotran (aiEDU):I do a lot of work. You know engaging funders and like school district leaders and we write. You know mous, and what we found is ai is really bad at writing. You know a personalized email. You can use it so you can say hey, I want to write a personalized email to zach kennelly, or kennelly, um, he's a teacher at dsst. And he'll say like salutation, zach, you know I'm writing on behalf of the ai education project, a non-profit. And it'll say like salutation, zach, you know I'm writing on behalf of the AI education project, a nonprofit. And it's like, you know, like I could spend time saying, oh well, I don't know, zach actually knows who we are and I want it to be like, by the time you go through that process, it's faster for me to just write an email.
Alex Kotran (aiEDU):But if I'm writing an MOU and I just need a, you know, a succinct section describing our work and how it aligns to, uh, this strategy that the district has, and you sort of upload the strategy that can be, that can be really effective, right and like.
Alex Kotran (aiEDU):The same goes for like fundraising and like writing grant applications. Um, if any of our funders are listening, like I hate to tell you, but like a lot of our grant applications are, um, you know, heavily influenced by ai, but we have not just a human in the loop but a human in the driver's seat, like holding the steering wheel with both hands, and so that means, you know, sure, using it to get a draft and to sort of like get past the routine aspects of the writing and content creation and getting focused on how do we make sure this is in the voice of our organization, that's really truly aligned with the goals that the funder has. Um, but you know, writing a overview boilerplate section at the top of the grant, that's not really and funders don't actually care about that right, like they're looking for something that is, you know, meeting their needs in terms of, like, clearly describing the work.
Alex Kotran (aiEDU):Um, so what do you? What like? What keeps you up at night in terms of there's a lot that keeps people up at night with ai, but in terms of, specifically, like your students and like you've seen a lot of amazing stories of that are inspiring, um, but are there any things that maybe people aren't talking about as much, that you think need to be on educators minds as they sort of start to use these tools and become more familiar with them?
Zach Kennelly:Definitely. So just about I'm going to like tie these two together. And the first is so like early I built a prompting framework like RCCP method role, context, content, prompt. I've like turned that into RCCP, which is role, context and content together in prompt. I've like turned that into RCP, which is role, context and content together, and prompt, and like it will change with reasoning models and things like that. But I do think, like you said, prompting is going to continue to keep people at the front end, and so it's like it's an incredibly valuable thing to learn. And the reason I say that is because at some point there's going to be real trade-offs Either you're going to be really good at prompting or you're going to have to give up your privacy to use AI, and, like I do think AI will get better at that at just intuiting things out but people will have to give up quite a bit of privacy to get that. And so this is, I would encourage people, just like you, to start prompting, start using it. I love the area of expertise. The next is I don't think anybody is getting very much from AI unless they're using custom bots. Right On Claude, this is project Gemini gems custom GPTs.
Zach Kennelly:I had this great experience today. Just like you said, a grant application I was planning a professional development and, like a year ago, I had built a professional development bot, trained on massive amount of like DSST, strategic plan, dsst values, a bunch of projects I'd done. I was able to go in give a brief set of context and had this. I was able to go in give a brief set of context and had this amazing professional development in like 15 minutes. It was exactly what I wanted. You know, I changed some of the case studies, et cetera.
Zach Kennelly:The power for so many people is like first learning to prompt, then understanding how to scale that through a custom, a custom project. So that's the big thing. What keeps me up at night is the accelerating inequality that is such a huge risk here that people who are really effective at leveraging ai, companies etc. Are going to be able to generate massive amounts of value, even if it's like, even if they're spending the same amount of time, they're creating more and better in that amount of time, and the people who are not just become consumers, who are entertained by it and are not capturing value. And so the acceleration of inequality in the age of AI is one of the biggest concerns that I have Acceleration of inequality. In a time of accelerating inequality and hopefully it can be a time of accelerating equality and that is what keeps me up at night and a big part of my mission and why I'm a public school educator, similar to y'all. I love your focus on rural and bringing people together.
Alex Kotran (aiEDU):I love your focus on rural and bringing people together is like how do we ensure that we don't leave more people behind with this extraordinary opportunity to democratize, or of many? I know that you're embarking on sort of a new chapter in your own career and, you know, once you've had some time to get your sea legs, I'd love to, you know, check back in and kind of hear your perspective. You know, from sort of like a slightly higher vantage point, but you know this really resonates. You know, if if we're going to assign more writing, have students write five pages instead of one because they have access to a computer, if a student doesn't have access to a computer, they're going to fall behind, and I think the same goes. You might say, okay, we're just going to increase the volume or make the project more complicated or challenging.
Alex Kotran (aiEDU):Knowing that students are using AI, there's this assumption built in that there's sort of equal access and that which is why this is so it's all connected. You know school is thinking about how do we make sure students have access to AI tools and have the privacy and the safety policies in place? Those are connected to teaching and learning. You know the questions around teaching and learning as well, zach Cannelli, thank you so much for joining us. I know this is a Friday. You're probably getting ready for the weekend. You're one hour ahead of me, so hopefully you have some fun plans in store. But it was really fun hanging out with you in San Diego and I'm looking forward to the next of many conversations with you.
Zach Kennelly:Same Alex Cotran, aiedu man so excited about this relationship. We're really grateful right for the opportunity to think deeply and have this conversation and be a part of the movement to empower folks leveraging AI in thoughtful, responsible ways.