In this episode, technology leaders discuss the integration of artificial intelligence in K-12 education with educator Douglas Kiang. The conversation examines essential skills for the AI era—including design thinking, close reading, and version control via GitHub—alongside data privacy considerations, the concept of distant writing, and evolving academic assessment models.

Douglas Kiang's Official Blog — The speaker's primary digital publication repository covering classroom workflows, design thinking, and the "AI Forward" series.

AP Computer Science Principles (College Board) — The advanced secondary curriculum framework for which Douglas Kiang served as a materials developer and independent project assessment architect.

GitHub — A software repository and version control platform integrated into the Menlo School curriculum to manage student multi-branching project workflows.

Google Workspace for Education — A secure productivity environment leveraged by school technology departments to keep user data in-house and isolated from external AI training datasets.

Perplexity AI — An information discovery and conversational search engine provided to high school students to enable multi-engine comparative analysis.

Google NotebookLM — A personalized AI research assistant noted for its widespread popularity among students evaluating internal course source documents.

ResearchGate: Distant Writing Paper — The scholarly publication by Professor Luciano Floridi of Yale University detailing the transition of authors into design directors.

Welcome to Talking Technology with Atlas, the show

that plugs you into the important topics and trends for

technology leaders all through a unique independent school lens.

We'll hear stories from technology directors and other

special guests from the independent school community and

provide you with focused learning and deep dive topics.

Welcome at everyone. I am Peter Frank, the Senior Director of

Certification and Operations at the Association of Technology

Leaders in Independent Schools, and I am Bill Stites, the

Director of Technology at Montclair Kimberly Academy in

Montclair, New Jersey,

and I'm Hiram Cuevas, the Director of

Information Systems and Academic Technology at St. Christopher

School in Richmond, Virginia.

Well, happy summer, guys. Howdy, do

happy summer.

It's been a while. What's going on? How are you

managing? It's funny because everyone thinks, oh, summer

hits, everything gets quiet, and everything kind of like, right,

ratchets down, and not in it. No, no, no, no, no, no, no, no,

no. Everything ratchets right back up. All the construction

starts. All the camps are getting going. Everyone thinks

that you have absolutely nothing to do in the summer, decides to

come out of the woodwork and throw everything that they had

pent up from the year at you all at once, because who's busy in

the summer? Not it, not at all.

Yeah, no. not to mention, you know, when people

are at conferences and they start saying, "Hey, have you

seen this? Have you talked about this? Can we think about this?

But actually, my spring and summer has been really the

spring and summer of Stites.

Oh,

it seems like since April I have been at some

event every month with mr. Bill Stites.

I am a luckier man for it,

and it's going to continue it through July.

I'm very conflicted about that, Hiram. I might go a

little crazy from that much time.

I'm sure you would, Peter. You most definitely

would.

Yeah,

you got time with me out in Salt Lake a little

earlier, a few weeks ago. I think you've probably had

enough. You're like, how much more do I have to endure?

Nah, it was great. It was great. Well, speaking of

Endure, as many of our listeners know, Atlas is going through a

significant transition, and guys, there are exciting things

coming for Atlas here, and with the departure of Atlas' former

CEO and host of this show, Christina Llewellyn. We're

revisiting all things Atlas, including our podcast, and

things like how often are we going to drop new episodes,

who's going to run the show here, and so on. We're working

all that out this summer, and in the meantime, what our listeners

cannot see because they're in their car or exercising, or what

have you, listening to this podcast, in addition to today's

guests, there's another person here. We all agreed we'd like to

have more voices on this podcast. We'd like to have more

of the Atlas community represented here, and we have so

many amazing, talented volunteers in our community. And

so we're recording all new episodes this summer, and we're

experimenting with guest hosts, every so often we've had a guest

on the show, when afterwards we all kind of look at each other

like they were a great guest, like they make a great host.

We're embracing that, and it is my pleasure to introduce to our

listeners our first of three guest hosts. She's a big-time

Atlas volunteer, the wonderful Lauren Merrill. How are you,

Lauren?

I'm well, thank you so much, Peter. I'm excited

to be here. As many of you all know, I'm Lauren Merrill. I am

the Director of Technology and Innovative Learning at Preston

Hollow Presbyterian School in Dallas, Texas. I am also proudly

displaying on screen my T-LIS certification. I'm a part of the

certification council, so that has been some of my favorite

work. I have worked in educational technology and

technology leadership for the past 20 years, mostly in Dallas,

Texas, and the opportunity to work with Atlas and go through

the original group of ECAD back in the day, before it was the

Ali Institute, and now getting to be on the certification

council for Telos, that's some of my proudest work, and getting

to contribute to this community, but getting to work at my school

and work with K through sixth grade students in a learning

difference environment has been a really unique part of my

career, and that's a really special thing that is unique to

the school that I work at now, and we're expanding through

eighth grade, which is also a really interesting path to be

on, and when I came on the podcast last, I had just

completed the major move into our new campus after a nine

month construction and move project from our old campus, so

that was a really intense thing, but I'm very happy that that's

not what I'm doing this summer. So we are settled, we are home,

and I am not moving schools.

Nice,

proud, and happy to be here.

Great. Well, again, we are very happy to have you

here. So I am going to get out of the way. A and Lauren, the

show is yours. Why don't you introduce our guest and take it

away?

Yeah, so we are so excited to introduce Douglas

Kyung, and we can't wait to hear more about your illustrious

career in both educational technology and all of your work

in computer science and AI, and just reading all of the work

that you have done is so impressive, and I think this is

going to be a really interesting conversation for our community.

I know that you've already stepped in and introduced

yourself to some of the Atlas community through some of the AI

work this summer through the trainings, but we're so happy to

have you on the podcast, and would love for you to just give

us a full introduction for people that maybe don't know you

yet.

Thank you so much, and it is such an honor to

be on this Atlas podcast. Thank you so much for inviting me. So,

my name is Douglas Key Young. I've been teaching for about 34

years now, in all different age levels and grade levels. I

started actually teaching elementary school. I currently

teach computer science. I taught an English course last year, and

I teach an AI literacy course, and I do AIPD for teachers at

Menlo School, which is in Atherton, California. And I've

had a really stimulating year, just done a lot of crazy things

with AI, made some mistakes, learned a lot from them, and

just love to share more about some of the things that we've

been trying out and hear from other people who are trying

things too.

One of the questions that I have is

thinking about those essential skills as we're looking at AI,

especially in the K through 12 space. What are you seeing in

the work that you've done, especially with your interesting

intersection between teaching computer science, but also

getting to teach English and literacy to students? What are

you seeing as those essential skills that we should be

teaching to our kids right now?

We've been working with AI and students in

lots of different kinds of subjects and I've been fortunate

to be able to teach in an English classroom with our chair

of the English department, I've worked with my computer science

colleagues and seen kids who are using AI to vibe code and create

all different kinds of things, and what I've been looking for

are the common skills that I think kids are going to need in

the AI age, regardless of what subject or discipline they're

using, and I think from my own perspective, what I think kids

are going to need is number one, I think is design thinking, I

think they need to understand in a larger sense systems thinking,

so understanding the design of the things that they build,

making sure that what they design is actually solving a

real problem for real people, because I think the challenge is

no longer going to be making stuff, it's going to be making

stuff that's actually useful and serves a purpose. The second

skill I think kids need more than anything else is probably

reading, so much more of the output from AI is written and

kids need to be able to close read and maybe put the same

prompt into different engines and be able to compare what

they're getting from all of not just for accuracy but also

reading for tone and reading for understanding and I think

reading is a skill that even our kids could get better at and

it's something that has never been more important than now.

And the third skill, which I think is kind of a niche skill,

but it's something that we computer science folks have

known for a while, is version control, because in a world

where it's so easy to generate content, I expect that a typical

workflow is going to be a multi-branching workflow where

you just try out different things and say, well, what if we

did it this way, what if we did it this way, what if we did it

this way, and then have a way to easily compare those branches.

Works the same way with writing papers, for example, usually by

the time you come up with a main point in your thesis, you're

kind of committed, because by the time you put in all the

sweat equity of building it out, you don't necessarily prototype

thesis, but I expect that that will change if people are using

AI to do much of the drafting, they're doing the planning,

they're working with AI. There's going to be a need to have a

multi-branching flow, and version history doesn't quite

cut it. So I tell my kids this is a great opportunity for a new

product, create something that's like GitHub, but for the masses,

something that provides ways to branch out. Right now, GitHub

does the trick. We actually used GitHub in our English class.

Right now, in our AI literacy course, every freshman gets a

GitHub account and training on how to use it. And we went

through the entire year. This year, next year, we're teaching

it again. So, by the end of next year, over half the school will

have a GitHub account and training and using it to do

version control, so those three skills - design thinking,

reading, version control - those are the three things that I

think are very important, and is what we built our whole AI

literacy course around.

Douglas, there's so much to unpack there, so the

timing of your comment. About vibe coding could not be any

more appropriate. I received an email yesterday from a colleague

who's at a conference where they're talking about vibe

coding, and he wanted to figure out a way to have access to App

Script, and as I started thinking about that request, I

started going down the cybersecurity rabbit hole of

what that means, and of course, one of the first people I ask is

Bill. I also asked some other Atlas members, like Shandor

Simon and Nicholas Bar Casey, who are really savvy in these

spaces, and I'm so grateful that you're here on this, because

maybe you can address some of the challenges that schools may

not recognize in terms of allowing vibe coding to occur in

an unsafe environment within a Google domain, which I think is

something that we can all understand, because if you look

at how we treated ed tech applications during Covid, we

all just started chasing, oh, this is free, we'll use this,

we'll use this, but now we've tightened the reins with many

tools, and Apps Script is just yet another tool that happens to

be part of the Google domain, but what does that mean for the

security of data in our schools?

I think the ramifications are huge, and I

think we need to be spending a lot of time thinking about

privacy protection of data, where the student records are

being stored and where they're held. Very, very important. We

have a great IT department. Our IT department actually is really

standardized on Google Workspace for education. None of the data

is used for training, and everything is stored in house.

And so my work primarily is with kids, and in fact, I'm doing

coding with AI course next year. I'm offering a new elective in

our computer science department. It doesn't have a prerequisite,

so you don't need to know how to code, but I'm going to have a

mix of kids, some who know how to code, some who don't. But

we're basically going to make stuff, and I need to teach them

that there are certain gradations of what you're

coding, you know. If you get to the point where you start

storing user records and user data, then you are bound to a

different level of security than if you're creating a web app

that stores everything locally as a cookie, for example, on

your own browser. No one else is accessing it. Frankly, a lot of

the things that I'm making for myself are really just

one-purpose apps, and that's what vibe coding has done. It's

really shifted the conversation. I used to teach an app design

course, and it always started with design thinking. Define

your audience, and we've always thought of apps as something

that serves not just your purpose but solves a problem

that lots of people have, we've always been looking at defining

the audience, but with vibe coding, the audience could be an

audience of one. You could create a tool or utility that

only you will use. From that flows a lot of other design

decisions. You don't necessarily need to build in settings, for

example, to do preferences if you're not going to have other

users using it. It's a very interesting time that we're in

right now, where I can create an app that I'm going to use for

one class, and kids can create apps that basically help them

learn in the ways that they best learn, and they're not planning

to share the app with anyone else, they're not caching

anybody else's information, and so the concerns are different

when they design them, so a lot of it is really just sort of

defining what are the different kinds of apps kids are creating,

and always paying attention to what are the questions around

personal data security and privacy that we need to attend

to.

So, hearing that your tech department is kind of

bought in on the Google Workspace for domain pieces is

one of those things that I take heart in, because it was a

decision that we made here at MK as well, and in particular for

the point that you brought up, which was that Google has made

it clear that they're not using any of that data to train the

AI, and then it's made it available across various age

levels, which was initially not where they were, so by removing

that age restriction by having that baseline protection with

regard to training, Google has been a place where we have begun

to direct efforts and in support and training and working

specifically in that area. You said something earlier, and I

want to thank you for what you did for Atlas with the AI summer

workshops, I was one of the participants in that and got a

lot out of it, but you had said that one of the things was going

to be this asking a different model the questions that you

were bringing up, and I'm wondering how you balance that

out, because one of the big questions that we have is what

are the approved tools like what are we saying you can use and

what are we saying you can't use and when you want to allow a

degree of flexibility with putting something into Gemini,

putting something into Claude, something into Chat GPT, but if

those aren't approved. How are you navigating that piece of it?

We've had conversations here. I've got my thoughts with my

school hat on, and then I've got my thoughts in terms of

personally how we could work around that, but the two don't

really align. So I'm curious as to where you've landed on that,

and then what you see.

Well, we have really decided to level the

playing field by providing all students access to Gemini Pro,

and this past year we also gave them access to Perplexity Pro,

and that was it. We did not give them access to other tools

through the school. Now, of course, this being, you know,

where we are in Silicon Valley, a lot of these kids' parents

work for Anthropic or Open AI or different companies, so kids

have access to other models. We don't prevent them from

accessing those models or using those models. I think it would

not be effective if we attempted to prevent them from using

things that they basically have on their phones or on their

laptops, but increasingly the school is moving towards we're

getting rid of Canvas and going to Google Classroom. We're now

basically making sure that everybody has access to Gemini

Pro. All of my teaching materials and student-facing

instructions are now oriented towards Gemini. So, I'll just

say, open up Gemini now. The kids know, because I've told

them a couple of times in the beginning of the year, look, if

you want to use something else, that's fine, unless I tell you

specifically, that's okay. But as a teacher, I have a fair

amount of control over what I promote. We're not a Mac school,

technically. We don't tell kids what to buy, but 99% of the kids

have Macs, because all of our teaching materials refer to Mac

stuff. We have steered it generally in that direction. I

expect we'll do the same thing with a tool like Gemini Pro, and

frankly, Gemini Pro and Notebook LM, the kids talk about it, they

use it, it's been very, very popular. So, no, we don't

prevent them from using other models, but in practice, kids

are tending to converge on the ones that we want them to use.

So, to that point, I think it's important, and I'm

curious as to how you talk to your faculty or employees about

this. Is because I think there is a real distinction between

students using it with their data, and then employees, and I

use the word employees, not just faculty, because my concern is

what all areas of the school are using it for, you know,

development office, admissions, business, those types of things,

is that while we may say, all right, a student can pick and

choose and use their own, their data is largely theirs, you

know. You can talk to them about what you provide into these

models. When we start talking about it from an employee

perspective, the data largely is not ours. It's data that is

school data, it's student data, it could be financial, it could

be all sorts of things. When you have those conversations about

choice and model, have you said to your employees, we're here

and you need to be here for x, y, and z reasons, or are you a

little more free form there and just giving like guard rails

around, we want you here, but if you're going to go here, you

need to follow these rules. Where do you land on that?

Right, so I am not part of the IT leadership, I

am a teacher, I'm not part of admin, so I can only tell you

from the perspective of being someone who's been told by admin

there are very specific places where you can and cannot put

student data, and we are being encouraged to keep everything

within Google Workspace, because it is no more or less private

than any of the other student records we keep in Google Drive,

and I know that there are certain more sensitive

applications, like development things like that, for which they

may be using localized models or other kinds of solutions that

protect the data even more, not 100% sure what they're doing in

those areas, so my experience has mostly been in the classroom

with my students, and with that, I believe we are a bit more

open.

You have a newsletter beyond the prompt,

and I was able to kind of read and subscribe because it's

really exciting in what you're talking about, and your

narrative is really wonderful. You have a theory, I guess you

could say, on close writing. You talk about close writing versus

distant writing, and that connection between close

reading. I just be curious if you could talk a little bit

about that, and where that falls into this AI conversation, and

the impact you've seen in either your English classes or maybe in

some of your AI classes that you're instructing.

Yeah, absolutely, it's actually something that I

kind of fell into while I was writing that trashy romance

novel. Our principal had said spend 10 hours experimenting

with AI and create stuff, and I'm a writer and a creator, so I

decided to write a trashy romance novel where I came up

with the story and the characters and the story beats,

and then I would have Chat GPT at that time write many of the

chapters, and in some cases I didn't like how it wrote, so I

asked it to fix it, and in some cases I just said I can write

this better, I just wrote it myself. After that, I wrote a

science fiction novel, then I wrote a steampunk heist

thriller, I wrote a courtroom drama. You know, and I say I

wrote all of these. I worked with AI. AI did much of the

drafting. In some cases, I used different models, and then

compared which scenes I liked better, and then chose those.

But these came out to be 450 1000 word novels, and for me it

was not writing. It was not creative writing in the

traditional sense. It was not close to writing in the sense

where we're scrutinizing how to write, how to write each word,

how to write each sentence. It was not that there had to be a

term to call what I was doing, and I came across a paper by

Luciano Floridi from the Center for Digital Ethics at Yale

University, and he coined this term, distant writing, and the

idea behind distant writing is you are not writing each word,

but you are directing, because if you think about it, during

the Oscars, you know, we give a best director award, but the

best director is not the person who was behind the camera,

usually the best director is the person who edited everything

together and looked at Barry's takes of the same scene, had the

vision in their head, and you know, if they make all the right

calls with editing, then they win best director, and so I

think students are going to become increasingly like

directors, where they're going to need to exercise their

judgment and their close reading skills to look at various drafts

and compare them, and if they learn to do a good job, I think

some students will be better directors than others. Then they

will be more or less effective at working with AI. So that's

this idea of distant writing, but one of the things that

Fleury talks about is that distant writing only works if

you have good close reading and close writing skills, so the

things that we have traditionally focused on -

reading carefully, word choice, sentence structure - those

become even more important in an era of AI and distant writing.

But I don't think that we can just focus solely on those

anymore, because the reality of the workplace is going to be

that I think the timeframes are going to become compressed for

how long people expect someone to create something written. I

think people are going to start assuming that you're using AI,

whether you used it or not. And I think we need to find where

that space comes from in our curriculum. I don't actually

have the answer to that. I don't know how we cover this

additional skill in addition to what we're already doing, but I

think the ratio of what we teach does change.

That's really interesting.

If you were to publish those, how would you

list the author of that book? Would that be you? Would that be

you? Would you acknowledge the use of AI in that? That's the

same thing we ask of our students in terms of what

they're going to be doing as well.

I think I would probably have to have a

disclaimer, at least for now, talking about how AI was used in

terms of either drafting or outlining or composing, but I

think the point I would make is that the reason I wrote those

books was not to publish. There are many other reasons to write.

I wrote because I had these stories that I wanted to get out

on the page, and I enjoy the heck out of reading multiple

versions of these scenes that I had drafted. I usually read a

lot of books, you know. For the past six months, I've hardly

read any books that other people have published, because I've

spent so much time pleasure reading stuff that AI has

drafted according to my design. It's almost like saying, tell me

a story, and then it this infinite storyteller tells you a

story, so it's a new form of entertainment for me. I think

it's something more than just pleasure reading, because it

feels more personal. They're my own stories, but you know, I

think again we're going to have to come up with new terms for a

lot of these kinds of things, and I will say that in the

romance genre, that is the one area of creative fiction where

people seem the most accommodating of AI-drafted

content. People don't seem to care, because romance novels

follow a very specific format. People actually don't like it if

you deviate from that format. They're very formulaic, but

people read them for a particular reason. They read

them because they like them. They don't particularly seem to

care what the provenance of that writing came from, if it's a

good start,

that's really interesting, and I think about,

you know, research skills and the ability to do a good query

was always kind of the beginning of how we use the internet, and

telling kids, like, if you can ask a really good question, you

can actually get to the answer you're seeking, and so now we

have this opportunity to go even further, where it's like, no,

that wasn't my question, this was my actual question, and

you're getting to go through the editing process in such a

different way, but working in the elementary setting and

working with kids with learning differences, something caught me

in this piece that you wrote about the student that always

struggled to get the words out of them and onto the page, and

the director role was actually really beneficial to them, and I

think about kids with learning differences too. Where these

tools will be really interesting for them, of how do you now take

a child that can't articulate their ideas by writing because

of dysgraphia or something like that, and now they have a tool

that can maybe intercede in a new way. And so we're still so

new to all of this, and I think that Bill's question about the

moral obligation to give credit to something that's not ours,

it'll be really interesting to watch the landscape as it shifts

and as it changes, but I think that challenge of what do we do

in the classroom, what are the ways that we can shift the

classroom space, especially in an English class, where there's

this beholden mindset that the kid must write the paper for it

to be their own, but your exercise is really interesting

too. So, what is your philosophy, and how are you

integrating that into the high school English class?

Well, I think it's becoming increasingly

difficult to come up with a piece of writing that you can

say was entirely 100% me and not influenced by anything else,

right?

And the question of ownership is an interesting

one. I would say probably the majority of the kids in my

class, we surveyed them, felt ownership over the stories where

they themselves worked completely with minimal use of

AI, but there were a few kids in the class who said that the

ownership of the story had less to do with the words on the page

and more to do with how closely did what was on the page reflect

the story that I wanted to tell, and I think that's going to be

an interesting question that we need to sit with. The truth of

the matter is, with Grammarly and all of these integrated AI

aids, quite often even a paper that a kid says they wrote

completely by themselves, there were turns of phrase, maybe in

the brainstorming process, there are places where AI kind of had

a hand in suggesting something or making something better, or

proving it, or clarifying their language. So, where that slider

goes, it's hard to tell. I think right now we're in an age where

people expect you to disclose if you used AI, but I think we are

rapidly moving to a place where you won't need to disclose that

anymore. In fact, maybe what people will need to disclose is

the fact that this was completely human-generated, and

there may be a cache, a certain value associated to something

almost like the way a craftsman or an artisan would say this was

completely handmade, but for the bulk of the writing, for the

bulk of the writing that's done in what we consume for both

entertainment and for business, it's going to be AI-generated,

at least in part to the extent that it won't be extraordinary

anymore, and in fact people will just judge it on the quality of

the words on the page and how closely it matches what you want

it to say. So I think this question of ownership is rapidly

going to shift, and I think we're kind of in a big

transition phase right now

based on the skills that you're talking about, and

your history as an early childhood educator. At what

level do you see the right time to begin to introduce these

skills at earlier levels? Like, have you thought about where

that marker is? Is it on a graduating slope, in terms of

how you would think about having these conversations, and really

starting at what level,

I think what informs me is my background as a

computer science person. I mean, we realized that far more

important than the language or the syntax were the concepts,

and so I worked with elementary school teachers to teach

unplugged activities, so nothing to do with the computer, but

having kids moving their bodies and working through activities

that taught them about key concepts like abstraction or

iteration, the concepts I think are the most important ones, and

so if we say that the important concept here is perhaps one of

top-down organization, if you're going to build anything with AI,

you need to have a structural idea of delegation. How are the

different parts going to work with each other? We can teach

that at the elementary school levels through activities

involving cooperation and role play without actually sitting in

front of a computer and using AI. I think if we abstract those

particular skills that we know are going to be important, if we

want kids to be directors once they get to the upper school

level, and they have those skills to be a director, and

then I think that's probably a more effective way to do it than

to have kids trying to direct AI at very early ages, because a

director only knows the difference between a good shot

and a bad shot, because most directors were actually behind

the camera in the day, they actually had to frame up a shot

themselves, so that they can know how to communicate with the

film crew. Now they're not the film crew anymore, but they were

at one point. So I think overall I would lean more towards using

less AI in the earlier grades, and maybe leaning into the

concepts, but the kids have to develop the foundational skills,

they have to learn how to read, they have to learn how to write,

necessarily, they have to learn how to code, necessarily. A lot

of that has to take place without AI, because if you have

AI doing the work, you're not going to learn it. And then you

get to the place where you know, last year, when I taught the

English course, we had mostly juniors and seniors, and they

were able to do an incredible. About using AI as a

collaborative partner and as a co-drafter, because those kids,

for the most part, had developed their skills of reading and

writing in an age that was pre-AI. So, our challenge is

going to be, how do we find ways to really get kids to stick to

building those skills first, because we know that you will be

judged by your ability to use AI, and your ability to use AI

will itself be influenced by how much you can do those skills

without AI.

So, Douglas, this actually, I think, a really good

transition, especially with your work with the College Board and

developing some of the College Board materials with the

computer science curriculum in mind. Where is that particular

test in this conversation? I'm not familiar with that

particular AP test. Should students be able to do close

writing of code at the AP level, or is the AP exam now designed

so that they have access to AI tools in order to produce set

product.

Both actually the students are required to learn

how to write code, but the College Board doesn't

differentiate between what language they use. So kids could

use a block-based coding language, they could use Python,

they could use Java. The idea is they need to master the concepts

of abstraction, iteration, conditionals. They need to know

what that looks like in the language that they're studying,

and so the mechanics of it. That's a good portion of the

exam. It tests that ability. Now, a big part of computer

science principles is also understanding technology's

effect on society, both the benefits and the harms, and of

course, AI falls squarely into that category. The effects of AI

on hiring practices, the implicit bias that's built into

AI, the harm that can be caused by deep fakes, and you know,

there are a lot of very current things that are part of that

course as well, but one of the things that I find most

interesting, and this just changed in the last few years,

is the kids have to do an independent project at the end

and they are told they can use AI if they like to help in that

project, but they submit their code, and then when they sit for

the exam, they have half an hour to answer a question about their

project, about their code, that they don't know in advance what

that question is going to be, and so you can immediately tell

the case where kids create something that's very

sophisticated because they just basically gave it to ChatGPT and

said build this thing, but then if you ask them some basic

questions, like if you change this program in this way, what

would happen? They have no ability to articulate that, and

so they don't do well on the test. To me, I thought that was

a very forward-thinking approach, because it says, look,

it's not about the tool, it's more about the understanding,

you know, go ahead, use the tools, but if you use the tools

in such a way that it harms your learning, then

you will not be rewarded for that. It's almost

as if this is a great blueprint to Bill's earlier question about

how early should you use it. It really is following the tenets

of what you've been describing. You've got to have that

foundation, you have to have those other skills and

understand those concepts first before you can actually do that

code

right.

It's also very interesting. I heard a great

example from, I forget, whether it was an English class or

history class, but it was the idea of a student will submit a

piece of writing, a paper that they may have written. The

teacher can use AI to create questions from that paper, and

those questions become the exam, or like the test of the student

based on what they wrote, and the student that has done the

writing the way in which you're describing will be able to

answer those. The one that has just completely fed it into AI

and submitted it won't, so the student knows that, all right, I

can submit this, I can put this in, knowing that I fully used AI

to do this, but when the assessment comes back to me, if

I haven't read what I've done, or if I don't truly know what it

is I've created here, whether it's a program, whether it's a

paper, whatever the output is, I'm going to be questioned about

it. If I can answer those questions, then I'm not going to

pass that assessment.

Yeah, I was kind of thinking even back to what

you talked about at the beginning of the three areas

that you feel are really important, and design thinking

was one that you listed as kind of a vital skill, and I think

that if we look at that from an early elementary age group and

start kids going through the design process early, those

skills then lead into better thinking structures around how

to use these tools later. Is that something that you're

seeing in those earlier age groups? And have you deployed

design thinking work with younger students as well?

Absolutely. And in fact, one of the most

effective activities that we do with kids is. A design thinking

activity where they're basically redesigning the gift giving

experience for another student, so it involves empathy. They

need to listen. What is it like to give a gift to somebody else?

What was difficult? What was the time when you gave a gift that

was really, really appreciated? And they listen, they take

notes, and then they prototype and iterate on a bunch of

solutions, and sometimes what they're designing is a process.

Sometimes kids decide to vibe code an app, but it's an

activity that really leans hard into empathy. And there are very

few times in this school day where somebody has to sit and

just listen to you talk for three to five minutes. It's

fairly unusual for kids to feel like the words that I say have

meaning and importance, and this person is listening to me, and

then making something that reflects back what they heard.

So, it's a great experience to be on the receiving end of

empathy, but also to have to try to develop those skills as well.

I am very curious. You've written about

something called the Unseen Unseen. Can you tell us more

about this?

So we're going through a WASC accreditation,

and I'm on one of many faculty committees. We were having a

meeting one afternoon, and we had basically put out a Padlet

to all the teachers, a big survey, and said talk about all

of these different aspects of the school, and so we had

basically taken this whole pad, but all of these things that

teachers had written, and we threw them into an LLM chat bot,

and we said, What are the five topics that resonate most with

teachers about our school? What is it that makes our school

special? And it came back with a list of five things, things

like, you know, our personalized block schedule, our faculty to

student ratio, you know, our college counseling. So, it came

up with a list of five things, and we all looked at them, and

they were all accurate. They were all things that we do, and

things that we're proud of. So, the room kind of nodded and

said, "Great, that looks great, let's move on. Saved us a lot of

time, but some teachers kind of raised their hand, and they

said, "Well, wait a second, what about the things that weren't on

that list, we don't know what was not included, because we

haven't looked at the Padlet report, as hundreds and hundreds

of pages here. But what if there was something that teachers put

in that should be on that list? We would be skipping over that,

and we weren't sure exactly why those were on lists. Was it

because the largest number of people said those things, or is

it because those are the things that make most schools special?

It is the thing that is the most probable, and so for that reason

it kind of rose to the top. We weren't sure, and I've started

to call this the unseen unseen. We don't know what AI doesn't

choose to show us. We've just been conditioned to check the

accuracy, because AI can hallucinate, and let's just make

sure that everything that it says is good, and I've seen this

happen with kids too. Kids will brainstorm ideas, and they'll

say, "Get five brainstorming ideas about my topic, and

they'll look at it, and they will be good ideas, but we don't

know if they're the best ideas. Maybe the best idea is some

half-finished idea they started at the dinner table with their

parents last night, but they just never had a chance to bring

it to the forefront. So, what I've started to ask kids to do

now is brainstorm on a piece of paper before you touch AI, spend

five minutes jotting down and brainstorming, and then ask AI,

and then judge. You can compare your ideas against the AI, and

maybe you still go with the AI ideas, that's okay, but you

don't have any way to judge it if you haven't given yourself a

chance to come up with the ideas first.

Yeah, that's really interesting, because I

think also we could end up in this societal place where we

take the status quo, or we kind of take the lowest common

denominator answers, and that could dampen the creativity, so

I think your point about what was said halfway at the dinner

table is really, really important, that we're pushing

creative thoughts still and not just taking those quick answers

from the AI tools. I think even Bill had brought up the concern

about not only how we're talking to students about using AI, how

students are engaging with AI, but how are all of the employees

in a school institution doing it? And I am sure that marketing

departments have started using AI more and building out any

kind of communication that's going out, and it's really

interesting, because I don't know if y'all have seen this,

but the more I look at other schools' social media, you can

tell who is just feeding it into chat versus who is actually

still trying to write it, or they've trained the voice, and

so I'd be curious about just like where do we go from here as

we think about schools as an institution, how do we tell our

stories and not become the same story across every school

because we're using the same AI tools to build the story.

I actually had an interesting experience. I

decided to try writing a new story. I had written these

novels by coming up with the plot and the character. Actors

and the story beats, but then I was having AI draft it, and I

was prompting AI to revise it until I was happy with it. Now

I'm in the midst of writing a new story, which started as an

activity that I was having the kids do an anthology of short

stories. Now I try to do a lot of the assignments that I give

my kids, so I started last semester, and I haven't stopped.

I continue to write, and I'm doing the writing myself, and I

flipped the roles. AI now takes a look at their finished draft

and gives me feedback, but it doesn't write for me. It

actually says, well, this could be a little sharper here, but

I've directed it not to tell me, so it's in fact doing the role

that I did before. And sometimes I'm writing multiple drafts of

the same scene, and having the AI compare them and say, which

one do you think is more powerful? And I've never had as

much fun. I feel more creative, and I feel like the writing that

I'm doing is truly my own, and it's my own voice, and it

definitely reads differently than any of the other stories

where the AI was drafting it, even though I had a hand in

revising those. So it's been a journey for me, and I think it's

been a necessary one that I've had to go through the spectrum

of having AI do most of one role and now flipping the role and

having me do something else. It's been a really revealing

look at how I write and how I create, and I think perhaps

there's something there. Perhaps we need to have enough

familiarity with AI to understand where is the balance

where I feel comfortable and where I feel I can be the most

creative and where I feel that I'm doing the best quality work.

We need to build a foundational skill so kids recognize quality

when they see it, but I am not ready to say that there is one

AI policy or one bright line that no one should cross,

because I think everybody is different as a creative, and I

think if there's one thing that we can agree on, which is we

need to feel proud of the work that we do, we need to feel

ownership over it, and it needs to look different than the AI

slop that's normally out there, and perhaps we give more agency

to the learners about the tools that they use and how they use

them. Perhaps that's a better way forward.

That's really great. And I do think that as we

work through this, it's all going to be a learning

experience. And I think these kinds of conversations and

getting to talk to folks that are in the classroom every day,

working with students, and then also those managing the data

security and the IT side of the house. I think that is where the

intersection is so important as we move forward, because having

somebody that's in there with students and hearing their

experience, their challenges, it's just so vital and valuable.

We'd love, as we kind of wrap up today, to know what you're doing

this summer. If there's anything that you are working on, or

anything that's exciting for you. It sounds like you're doing

a lot of fun writing, but are there any other things on the

horizon for you before the school year starts?

Yeah, I'm actually in the process of not

only finishing my creative anthology, but I'm in the

process of writing a book for teachers about the AI forward

classroom, so everything that we learned about getting kids

invested in the writing that they're doing, or the coding

that they're doing, it really is going to be intended for

teachers to teach all different kinds of subjects. I could not

do what I do without the collaboration that I have with

Michael Bada, who's the head of our IT department. He has been

tremendous as an ally and an advocate. I want to write as

well about the good work that can be done when teachers and it

collaborate and are on the same side in helping not only to

empower kids but also to protect them. I'm doing some traveling,

I'm going to Columbia. I've never been there before, but I

have a good buddy of mine who moved out there with his wife,

and so I'm gonna spend a weekend in Colombia visiting him, and

then we'll see our kids on the East Coast, so it'll be a full

summer, but I expect they'll return rested and rejuvenated

and ready to start another AI forward year.

That sounds pretty amazing. I'm hoping that

Hiram and Bill start on their zombie AI anthology sometime

soon for all Atlas listeners. We've

already started it.

I think it's going to be amazing. This has

been such a pleasure, Douglas. We're so glad you were able to

join us.

Thank you.

This has been Talking Technology with Atlas,

produced by the Association of Technology Leaders in

Independent Schools. For more information about Atlas and

Atlas membership, please visit the atlas.org If you enjoyed

this discussion, please subscribe, leave a review, and

share this podcast with your colleagues in the Independent

School community. Thank you for listening.

Thank