What AI can and can't do for teachers, with Ty Davis

Gautam Thapar: Welcome to In Practice. I'm Gautam Thapar. I spent my career in schools as a teacher, a principal, and a school system leader. And these days I'm obsessed with one question. What does AI actually mean for the day-to-day work of teaching and learning in K-12? I'm also the founder of EnlightenAI, but this show isn't about Enlighten. It's about teachers and school leaders doing the real work on the ground and how they're navigating the promises and the perils of AI. I'm thrilled to welcome my friend Ty Davis, CEO of ICEF Public Schools, a network of charter schools in South LA and Inglewood. Ty's resume is the kind that makes the rest of us blush a little. He went to Yale, then became a master teacher, after which he attended Harvard Law. After practicing law, he designed, founded, and led Vox Collegiate of Los Angeles for almost 10 years. Vox was named a California Distinguished School under his leadership. And if that weren't enough, this year he's a Broad Fellow, one of the most selective leadership programs for school system leaders in the country. But what I love most about Ty is that he has never stopped being a practitioner. Just a few weeks ago, when a teacher left, the CEO of ICEF walked into the classroom and taught class himself. That's who we have the privilege of learning from today. Ty, welcome to the show.

Ty Davis: Thanks for having me. Excited to chat with you today.

Gautam Thapar: Let's get right to it. Could you take me into your office for a minute? Talk to me about the biggest problem or problems that you're facing as CEO of ICEF.

Ty Davis: Oh man, if you could see all of the post-it notes on my computer screen right now, I could run through a list. But the one that keeps clawing at me, the work of schools and the role of schools in our society has broadened significantly in the last 30 years. And with that, the role of every individual within a school system has broadened and become more complex and more challenging. Finding people who are equipped to do a job that is nearly impossible by design is becoming harder and harder. And that's most acute in both teachers and school leaders. I think what we're asking people to do is to be martyrs in certain respects. We want demigods doing the work, but we are mere mortals. And so, one of my challenges is figuring out how to make the work that is so essential actually doable by more people. We have talented, really bright people, and I feel like too often what I'm asking them to do is setting them up to fail. So, I'm figuring out how to break it apart, which right now I don't have the answers, but I'm I'm hopeful that we'll begin to turn the corner on that relatively soon.

Gautam Thapar: Well, maybe by the end of the conversation we'll have some of those answers. Tell me a little bit more about that. What have you seen as being the change in teachers' roles? What are they being asked to do now versus maybe when you started teaching that is significantly different?

Ty Davis: Yeah, I mean, one is just how accessible the teacher is expected to be. The role that technology has played in all of our lives is like we can never untether from our work. You know, you can't untether from your boss if you have your email on your phone. If I'm a teacher with 30 kids in the classroom, I have 30 to 60 bosses who I'm in charge of taking care of the most precious thing in their lives. So, the expectations around like how our teachers communicate and bring families into the classroom and bring them along this journey, there was a perceived deference to teachers decades ago that doesn't exist anymore. And so, there's a lot more expectation for teachers to prove that what they are doing is the right work. And while it may only take 5 minutes to respond to a parent, the emotional and the intellectual lift of even preparing for that communication is a lot to ask of people. And to do that day in and day out gives an increased weight to the work. I'll also say that something we don't talk about enough is the testing regime. The tests are much harder than they were in the state tests from 2009 to 2026, completely different tests. And so, what we are expecting, the level of mastery that is expected for us to lead students to, like that is a greater lift on teachers. I'm not saying it's wrong, but the work and the way that we structure the work has not shifted to accommodate the additional rigorous lift that is expected. And then you lay on top of that what we know now about, you know, student social development and their wellness. And now, I as a teacher not only have to be a math content expert, I also have to understand child development in ways that I wasn't expected 2 years ago. It's tough. We don't need to get into everything that changed under COVID and distance learning with a lot of the new administrative requirements that are there. But, you know, there were things that we could kind of put off to the side and not prioritize in years past that are now immediate priorities along with preparing a student for the next grade and beyond.

Gautam Thapar: Yeah. I mean, I felt some of that when I was in the classroom and as a school leader and it feels like just across the board we are asking teachers and educators in general to do more with less even more than we already were, which was pretty extreme.

Ty Davis: Yeah.

Gautam Thapar: This seems like the job that AI is built to solve. It enables teachers to do more with less. So, to what extent are you seeing AI address or solve some of what you're bringing up?

Ty Davis: Yeah, I mean, to give teachers back a bit of time on grading and feedback. And then there's other areas of lesson planning I see some promise in of you know, course and assessment design and review I see some promise in. Where I'm I'm seeing a challenge is I've stepped into classrooms and seen AI supported materials but the teacher doesn't have the underlying expertise to realize that what was created actually isn't good for kids. Or it might be 70% of the way there but if you would have you still have to do that extra 30%. So, coaching teachers to get to a level of expertise where they can really use AI as an accelerant in their practice is going to be hard because as a novice teacher I feel like I'm drowning. I'm just looking for a life raft. And I think AI is becoming a life raft in some of the wrong areas. So, it requires organizations, school systems to help practitioners figure out like where is the appropriate use while we figure out some of the more nuanced pieces.

Gautam Thapar: You mentioned the last 30%. Talk to us about what that 30% is. What are you seeing in these materials? And what is that last 30% that you're seeing missing in either the materials or it sounds like if I'm reading into your answer, there's something in how the teacher is prepared around those materials.

Ty Davis: Yeah, oh man. We can get into a lot about intellectual prep for classrooms, but part of that last 30% is if you are designing curriculum materials for your classroom, you should actually be designing them for the students in your classroom. And so an AI created a handout for an ELA class may be asking a question that actually isn't aligned to the way you've been talking about this concept with your students. And part of that could be an issue with the LLM model and the way it's interpreting it, but part of that could just be like you as the teacher weren't specific enough in your prompting. But the I don't think teachers want me to put a burden on them of giving them like professional development in prompt engineering right now on top of everything else they have to do. So I don't I don't fault them for that, but if we are doing things that are supposed to help us teach kids, the products and the tools we develop actually have to be designed for our classrooms. And what happens I think in a lot of what is produced in these models is it's materials that are created for a generic or generalized classroom that don't fit our individual context, whether your students are accelerated or your students have some gaps and need additional scaffolding, you have to be the master of that fine-tuning of the materials that are being produced. And it doesn't mean that you can wait until the last minute to do this stuff. You still have to be organized and do the prep work. And then the other 30% is I don't just see this in the instruction side, I also see this on the administrative side. We have become very good at creating plans and SOPs and policies and documents. Who's taking the time to internalize those and know how those actually play out in practice? And that's where AI has made us feel very productive, but there's still a real gap in implementation, and implementation, that is the human work that we cannot skip. I do think AI gives us more time to think about and to actually do implementation work, but I think folks are and I'm also guilty of this and have been slowing down intentionally, we've been skipping the implementation work assuming that just creating that thing with AI is the work in and of itself, and it's not.

Gautam Thapar: Yeah, I must confess that there are times where I think back to compliance reports that I had to fill out that took a long time and I maybe get a little nostalgic for the idea that I could have used an AI to speed up that process, but what you're describing sounds like a microcosm of the broader problem that folks have with AI, which is it allows you to go faster, but that last 20% is where a lot of the value comes from. So, I guess my question for you is what do you think the net effect is having seen a pre-AI world and a post-AI world in schools? Do you think that it is improving teaching and learning? Is it about the same? Or is it possible that it's made it slightly worse?

Ty Davis: Yeah, I don't think I see it as a pre-AI versus post-AI. I think we are like in the midst of like very nascent phases with the new technology. And like this is just going to be messy. I don't and I don't get frustrated by the messiness. Like this is normal. This is how it should be. The initial implementation of common core was really, really messy. Anytime a state rolls out a new requirement, it's messy. We figure it out. I see this is as no different. My concern is people rushing to find a conclusion about what the implications of AI are going to be, whether that's you know, the AI optimist or the AI pessimist. We just don't know yet and that's okay. Now, what we should not do is rush to experiment on kids because every day that they're in a classroom is a day they don't get back. We as the adults get to do these days over again each year, the kids don't. So, we do need to protect the things that are already working and not rush to throw out the baby with the bathwater, but I think that this moment of opportunity and creativity is a chance for us to rethink and I think it's open doors to rethink instructional models in a way we haven't since we did distance learning. I think that I think distance learning opened the door for people to say, "Maybe there's a new way." AI is doing that again. And I also think we need to be honest about the fact that even pre-COVID the system was not working for most of our kids. We wax nostalgic about the good old days, but in the good old days, I was still an outlier amongst the community at my school, especially of young black men. There were a handful of us that got out. But, the fact that it wasn't the norm for all of us to have academic success and have opportunities to go to college means that the system was flawed. And so, we should be seeing where AI might be able to close some of those opportunity gaps that we absolutely know exist and many of us have committed our careers to closing.

Gautam Thapar: That is a perfect segue to talk a little bit more get your vision for what you would hope that change would be. It's a great point that right now we are in a transitional phase to paraphrase what you described. If you're thinking through what you want the ultimate impact of AI to be, the problems that it can and will solve in schools if we get this right. What does that look like? What problems are solved? How do your schools look different?

Ty Davis: A big one is freeing people up to do the work that only a human can do in teaching. Teaching is people development coaching work. And you know, it'd be great if we could set up an AI bot or a tutor to, you know, teach kids and guide them, but a computer can't give you that nudge or that pat on the back or slide you that snack or say, "Hey, hey, go splash some water on your face and come back in 5 minutes." It takes a human to develop the relationship that allows a student to work through the discomfort of learning. And I tell kids this all the time. What I'm asking you to do kind of sucks. This is hard. You shouldn't just want to do this hard thing. Like it is normal. The resistance you're giving me is normal. It is my job to help you work through it to get to the other side of it cuz on the other side of it is joy. But we can't escape the discomfort of the learning process. What AI can do is to give me more time to be better at guiding students through that discomfort. Whether it is instead of having to meticulously do line by line grading and feedback for an essay. You know, Enlighten allows us to create models based on our expertise. Even being at the difference between being able to dictate my feedback versus having to hand type it is a big saving of time. The ability to do data analysis and to challenge my thinking with a thought partner instead of having to wait for PLC to do it. The ability to potentially differentiate materials for students. Our classrooms, we talk about differentiation. You know, I walk into an eighth grade classroom. There's students reading on a fifth grade level, students reading on a college level. They both deserve to be met at their zone of proximal development. For me to do that as an individual is really hard. But for me to now have an AI assistant to help me do that makes what is seemingly impossible possible. And so that's where I see like AI truly being a This is a little non-PC, but I call it a performance enhancing drug. Especially for those of us with some level of underlying expertise, it allows us to get that additional bump and I think that's what we should be using it for. And when I think about young leaders coming to the profession, you know the experience of being a first year leader of the school, especially in the school founder. You are doing 13 different jobs at one time and probably doing none of them well. And it is incredibly lonely. Having an AI assistant, I'm having some folks test out Legit Leader, which is a platform out there. My principals can't always call me. Sometimes they can't call our chief of schools, she's busy. Who's there with you to help you actually process your thinking? While AI is not supposed to do the thinking for you, it is it can help you slow down. It can help you see your blind spots and challenge you in ways that no one else in your building can. And then allow you to be better at the human stuff, whether that's working with students, families, or even coaching a staff member through a difficult time. And so when I see examples of AI working and allowing people to be more confident and better at meeting the needs of students and families, that's where I'm just like, "Yeah, let's let's bring it on. Let's let's give this a chance." And it's because this thing is just so damn hard. We need to be open to help.

Gautam Thapar: Yeah, it seems like the challenge is to get this optimal balance of the teacher or the school leader and the AI so that the human is doing those things they're best at, but not as much as we're asking them to do in the current environment. I'm wondering if you could try to make that even more concrete for listeners. What might that look like in a day-to-day class period for a teacher? Is it that their materials are AI generated but with heavy editing? Does it mean that students need to be on AI programs during class and the teacher is circulating and facilitating more than they are in front of the room delivering core instruction? Or is it something else?

Ty Davis: Oh, gosh. They can be myriad things. One way I'm thinking of it is the students working in a classroom and not necessarily needing to wait for a teacher to know whether they're on the right track and to progress. And also allowing the teacher to go deeper with individual students who need more intensive support than others. Now, that could be them working on an AI platform. It could be them working on Google Docs and submitting it for feedback that the teacher has constructed an AI model with Enlighten or another system that is going to get feedback aligned to what they want to do. It could be AI if I'm a teacher and I'm creating an assignment for students, I know what the most likely misconceptions are going to be. What are now the off-ramps for students who hit those misconception points? So, that like those things are already built out and ready to go versus me having to now think of something and respond on the fly. And so I think it can create a classroom that is a little more free-flowing because you're using AI systems to do things like immediate feedback, answering questions, generating extension work, generating review work. And it all can be because like the models are now built, so we can have relatively closed systems. It can be built on like what we want to do as a school or as a classroom versus a generic LLM. And like that's where I see as a teacher feeling less tethered to the front of the room, less restricted by scripted curriculum, and more freed up to actually have one-on-one interaction and one-on-one time with students in the classroom, at the secondary level in particular.

Gautam Thapar: This makes me think heavily about the role of the teacher, and you and I have talked at great length about how difficult it is to recruit, hire, train, and then retain great teachers. So, I'm wondering does this vision of AI in the classroom require more skill, less skill, or about the same skills as teachers currently have? And is it a fundamentally different shape for what teachers are spending their time and energy on a day-to-day basis?

Ty Davis: What I'm thinking now is instead of more or less skill, I'm thinking of different skill and more potentially more focused skill development for teachers. To ask every teacher to be both a content expert and a pedagogical expert, I think it's asking a lot. Especially with the varied needs and varied expectations now that we're encountering in classrooms. And so I think you can have with AI, you can have a distributed expertise model where you might have a lead teacher or teacher coach who is really good at facilitating practice in the classroom. They are that teacher that is fantastic at relationships. They are the in the moment, I know how to respond and I can be dynamic and I can make sure students for this hour-long time, every student is working and focused and excited on what they're doing. Problem right now is we want a teacher to do that and prep and grade and continue to develop their own content knowledge. But what if that coaching in the classroom was what we really focused them to develop on? And then someone else was then doing the deep intellectual prep of the planning the materials, grading the materials, doing the deep content work. Maybe coming in once a week, you know, like a college professor or something and doing some in-depth content work with students. But in the other four days of their week, they actually have time for the planning, the feedback, the grading, the assessment and all of that. It might require some additional staffing, but I actually think it could be accomplished pretty easily. It requires a lot of material creation, which is where AI comes in. It requires a lot of individual attention to students, which is where AI comes in. So, you have this expert at student development and work facilitation and this expert on content able to do all of the varied work that is required to be an excellent teacher. But, instead of investing it in one person, potentially investing it in multiple people.

Gautam Thapar: That is a very compelling vision. No surprises here. As you think about where you and ICEF are in the change management process to get to that vision, how would you grade that? Where are you? Where do you want to be? What are you running into in pursuit of that change?

Ty Davis: Yeah, we are at the nascent stage. We're very early on. And a big part of my work is creating a sense of urgency around this and making sure that we are asking the right question and addressing the right problem. You know, I have one view from my seat and try to take perspective of the teacher. And like that's a big reason why I still will hop in the classroom is because I need to stay close to what the work actually is. But, I need to get out there and create urgency across the team from our operations managers to our instructional aides, to our teachers, to our principals that this is something worth addressing. Because what is going to be the real challenge, not just for ICEF, but for all of our schools, is as we go through this change, there is going to be a sense of loss. And that loss comes from like a genuine fear of what the unknown is. And we shouldn't dismiss that in people. It's our job as leaders to help guide them through it and to give them voice in what the change looks like. And so that's where I am right now is getting everyone to understand like this is the problem worth addressing and getting them comfortable with me leading the organization through it because they're you know, telling a teacher "Hey, we're going to have you be a learning coach." And there's going to be this other expert that comes in. I if you know, if I'm a 28-year-old teacher, my immediate thought is oh, you're replacing me or I'm getting a demotion. I need them to understand like I'm I value the learning coach the same level as I do the content expert. Y'all are just doing different work. These are different jobs and like my duty is to make sure you can be successful either way. This is not going to be easy. But like you know, the perceived downgrade is real. The perceived loss of control and autonomy is real. And so there's there's a lot of pre-work done to build trust, to build urgency. And so and to tell a consistent story that we can all address together.

Gautam Thapar: Well, on a note and to share some chicken soup for the system leaders' soul here, what are those conversations like with teachers, with leaders? I imagine that all layers of the organization there's a shift that's at play where folks are being asked to do different things. How do you manage the change steadily without too much disruption, but the right amount of disruption to get to the goal post?

Ty Davis: Oh man, that's an ever-changing balance. One week people are ready to go, the next week you got to dial it back a little bit because the change tolerance lessens. But one part of that is in some work we're doing right now is talking to folks about what do we actually want to promise the kids? Like can we all get aligned on this is what we are committed to do for every student that comes through our doors? Because if we can get alignment on that, then we can say we can better prioritize. We can better allocate our resources, not just money, but also our time and energy and be able to say no to certain things more easily. And so I'm being open to asking those questions and not assuming that I'll have all the answers. If it was me locked in a room, we would just do speech and debate and communication classes all day and algebra. Like that's all you need, but luckily I have people around me pushing me on the role of belonging and its centrality to our work at schools and arts and athletics and those being essential pillars to a well-rounded education. I have to be humble enough to just know that I don't have the answers, but collectively we as a team do and uncovering those will I think will unleash us to move forward with real ferocity.

Gautam Thapar: I want to shift gears just a little bit here and talk a little bit more about your personal use of AI and the bright spots that you see that maybe canaries in the coal mine for the vision that you're laying out. So in your personal work as a school leader, what does that look like on a day-to-day basis? How do you use it? Where is it helping most? Where do you see the limitations?

Ty Davis: Yeah, I use it consistently. A big one for me is as my thought partner. As I was, you know, redoing ICEF's strategic plan, I would have an idea in my head. I need someone to challenge me and check my assumptions there. And there was even, you know, ICEF, View Park High School, one of ICEF's schools, is on a percentage basis the blackest high school in the state of California. And so I was beginning to go through the process of thinking through, okay, like what does it mean to successfully educate black students? Like what is what are the models? What are the ways that we should go about this? How might we need to structure the school differently to be more effective? That actually working back and forth with an AI model actually got me to realize I was actually asking some of the wrong questions there. While I do think that race and culture and ethnicity matter and are things like we do need to be culturally responsive and culturally sustaining in the work that we do. What I was really addressing was the fact that the model of education we have right now generally is simply not resilient enough for our kids. And the kids who are impacted the most severely are our students of highest need, which for View Park are students who are 98% socioeconomically disadvantaged and 93% African American. If you look at the performance in schools, those students have the largest gap in performance. And so a system that has design flaws is going to harm them the most. It is not that I need to design a new model for black students, it's that the model in and of itself is broken. I would not have come to that and recognize that as a problem by myself. I think over time, talking to people, working through it, I would have gotten there. I was able to get through that in 2 hours back and forth. Me and ChatGPT at the kitchen table on a Sunday morning. And so like the ability to have somebody sitting at the table feel like I have someone sitting at the table with me checking my thoughts and my assumptions. That is how I use it most centrally. And then of course there's organizing emails, helping me do some administrative stuff that I absolutely despise, refining slide decks. My time should not be spent 4 hours designing a slide deck for meeting. I can do the thinking and then allow an AI assistant to do things like that. So that's been incredibly valuable to me to like get me back time for the hard work of thinking.

Gautam Thapar: That is an incredible story. And I sometimes find that these models just get me to write down my thinking in a way that I might otherwise keep things in my head and just that act, Regardless of what it says to me, can help me get to that better understanding. My guess is that most of those insights were in you somewhere, but being able to have a venue to think it through even with yourself probably made all the difference.

Ty Davis: 100%.

Gautam Thapar: You are a father of two and you know, let's say an unfortunate amount about how schools work. And you know quite a bit about AI. And so I'm wondering to what extent is it impacting what you're seeking in the schools that you're sending your kids to or generally what you want for them in their education?

Ty Davis: It influences me to choose schools that are going to develop I have two girls, Scotty and Blakely. To really support them on the human the social side of things understanding the world like super like as liberal artsy as can be. The technical aspects of things they can learn and figure out with technology with AI and all of that like they'll they'll learn it. What I think is harder to learn as you get older is how to use these systems and these technologies in the human context in a responsible and ethical way. So I want schools that are going to really give them the foundation of knowing people of respecting people of taking perspective of understanding their history so that you know as they are the future builders of technology they are doing that in incredibly responsible ways. And then you know I'm fortunate in that if there are gaps in other areas like my wife and I can very much like compensate at home for any gaps that we might find in a school system which is you know nice to have that privilege. But yeah we actually recently switched schools and it was not because the other school was not good at the human part but as we found a new school that was a big thing for us finding a school where she would both see people who look like her and come from her background but also be exposed to very different students. Particularly for me being in a an inclusive environment with students from various economic backgrounds and students with learning differences. I think being able to see the humanity of people who are different than you is a superpower as a human and allows you to be the best version of yourself. And so that's what I want for the girls.

Gautam Thapar: Well I can't think of a better place to pause and go to our lightning round at the end. A very compelling vision for school, not just for our own kids, but for all kids. So, quick lightning round. I'll ask you four questions and we'll see how lightning they really are as this is our first run-through. Number one, one piece of advice you'd give your former teacher self.

Ty Davis: Pack your lunches.

Gautam Thapar: One piece of advice to your former school system leader self. It can be pack your lunches also.

Ty Davis: No: find community with other leaders.

Gautam Thapar: Who is your favorite teacher growing up?

Ty Davis: There were so many. My earliest favorite teacher would be my second grade teacher, Mrs. White.

Gautam Thapar: Let's go non-lightning for a second. Say more.

Ty Davis: Mrs. White was this brilliant woman who made me believe that I could actually be the president of the United States one day. There was a poem on the wall and she did it every year with her classes, but ours was class of 2005. When we grow up, we'll be ready to strive to reach our goals, to set our pace, to make this world a better place. And we had to say it every single morning and she absolutely believed it in us and stayed a part of my life for a long time. She passed away 2 years ago now, or no, 1 year ago now. Just a beautiful, beautiful woman.

Gautam Thapar: An inspiring story. And for what it's worth, I also think you could be still president of the United States. Last question, how can listeners who heard any of this be helpful to you.

Ty Davis: Oh, man, continue to have conversations and challenge things. Challenge systems, because that's what gives us as school leaders, as practitioners, when you challenge things, that give us the freedom to now go out and figure out like what comes next and what can be better for kids. When we sit on our hands and accept the status quo, we get locked into things that entrenched systems that are harmful to a number of our students.

Gautam Thapar: Ty, thanks so much for being on the pod.

Ty Davis: Thanks for having me. Appreciate it.