How I used AI to create my seating charts

Why I used it, how I told students, and when I turned AI off to turn my teacher radar on

Students are back! I launched my classes this past Monday, and I have the privilege of supporting a new cohort of 99 juniors and seniors as they explore their futures.

(Seniors came in the first day wearing elementary school backpacks, part of a TikTok trend whose opacity to me makes me feel all 21 of my teaching years…)

With students and desks come seating charts.

If only the floor would stay that shiny until June…

I used AI tools to help me make this year’s seating charts. Here’s how and why.

Step 1: Create survey questions

I wanted to be even more intentional than usual this year with my seating charts, given the new abilities AI affords, so I started with a new step: creating a detailed survey for students to complete to inform my seating chart.

My class is about purpose- and career-exploration, so I gave Claude Opus 4.1 and GPT 5 Thinking this prompt and attached some relevant reference documents.

I run the Summit program. The program description is attached. You're an expert in adolescent purpose development and career connected learning. I just started the school year and I want to sort students into groups based on their career interests. Generate survey questions I can give students to determine their career interests and sort them into groups. They will be in these groups helping each other on their purpose and career exploration journeys. So I want the survey to generate the kind of information that will best help me make those groups. Specific career field interest of course, but also other kinds of more subtle information it will be good for me to know. You can ask open-ended, multiple choice, and Likert scale questions. Draw on all relevant best practices for this in your training data and on the web.

I got initial surveys from each model, Claude and GPT. I then gave this simple prompt to both models. (This is a best practice I learned from Ethan Mollick’s work.)

Make the survey better.

I then gave each LLM the other one’s survey draft and gave them both this prompt.

Does this survey include anything we don't but should? Revise our survey to include the best of both, our original and the new draft. Be candid.

I then compared the two new surveys and decided I liked GPT’s better.

I then went through and edited the survey, deleting some questions and revising others.

Step 2: Create the survey itself

From painful experience, I know how long it can take to make a complex Google Form survey by hand. And the survey I created this year was intentionally longer and more complicated than any I’d used before. That was the point of leveraging AI, to make something more effective and detailed than I would have time to do on my own.

So I then used AI to do something else I didn’t have time to do on my own: make the Google Form survey itself, from scratch.

First I wrote:

Turn the survey into a Google Form.

I got this in response:

I couldn’t download that file, so I had GPT write out the entire code in a format I could cut and paste. I then pasted the code into the script.google.com site (which I had no idea existed before this — I was an English major in college and then an English teacher for 18 years before launching Summit).

GPT had told me above in step #2 to choose “createForm” but I couldn’t figure out how to do that. I’d never used this site before. So I uploaded a screenshot and asked for help:

I then ran into a couple of code issues; each time I uploaded a screenshot and asked for help. Less than 5 minutes later, I had the right code running and next thing I knew I had a complete Google Form survey ready to post in Classroom.

Note: I uploaded the “Summit” image myself using the Google Form design button. GPT created everything else.

Step 3: Give students time to take the survey

The next day, I gave students 15 minutes in class to complete the survey. Before I did, I told them this:

I created a new survey this year to help me make the best possible groups for our class. As we discussed on Day 1, no one reaches a summit alone. We’re going to help each reach the life summits we each have in mind. Success requires community.

With that in mind, please be thoughtful and honest in your responses to this survey. Some of the questions are specifically about your life goals and future plans, and others are more general and get at how you like to work and collaborate best with others.

I’m giving us 15 minutes right now in class to get this done, since I know it’s longer than usual. That should be more than enough time; last period finished in about 10 minutes. I do need you to get this done now, though, so that I can make our groups and new seating chart tonight.

Students then took 10-15 minutes to finish. As they worked, I walked around the room to answer any questions.

(I also used this time to practice remembering everyone’s name in class, walking around the room and saying names in my head. I told the students on Monday that I would be giving myself a “name quiz” this Friday, the end of the first week. During the quiz I have to say everyone’s name in front of the class; they get to decide where I start and what order I go around the room. It’s hard to build community when it’s the second week of school and you don’t know everyone’s name, and they know you don’t. The students whose names you can’t remember are the ones who especially know.)

Step 4: Use the survey results to create groups

Later that day, I downloaded the survey response data. I first made sure it was anonymized. I replaced all student names with numbers and read through all the open-ended responses to ensure there was nothing personal or identifying in them.

I then uploaded the survey data for each class one by one to GPT, which does a better job with spreadsheets in my experience than Claude. (Though Claude is better at writing feedback for me.) I gave GPT this prompt.

Ok, here are the 6th period survey responses. Make groups based on this. For each group tell me the overall career connection between the students, the students' number, and explain why you put them together. No more than 5 students per group.

It created great groups, but it described them in an overly elaborate and confusing way. I needed something simpler to use while teaching so I gave it this prompt.

List each group again with just this for each group:
- Overall career connection
- Students' numbers
- Students' individual career interests within that overall career field
- Why you put those students together

That gave me a list I could easily reference during class.

Step 5: Turn off AI, turn on your teacher radar

At that point, I cut and pasted GPT’s group ideas into a Google Doc and closed the GPT tab. It was time for me to take over.

I looked at the students in each group and thought about who will work well together from what I’ve seen so far. I considered factors like student personality, academic history, and classroom dynamics.

I moved some students around based on that, keeping as much career alignment as possible but also being realistic. Students will only make progress towards their career goals — or really any classroom goals — if we put them in a position to succeed. In this case, a physical position.

Speaking of which, I then took the final step: figuring out which students would sit at which desks. (As you can see in the photo above, one easy way to do seating is to put a number on each desk before the first day. I write the numbers on post-it notes with a sharpie and then use packing tape to make sure the post-its stay put.)

I decided where in the room to put each group with the following in mind:

• Which students need to sit closer to the door for personal reasons.

• Which students should sit further from the door because they get distracted every time someone walks in or out.

• Which students have trouble seeing the board.

• Which students want to sit further from the front and will still be focused there.

• Which students need extra academic or personal support and so should be closer to the front.

• Which students can’t sit close to each other without getting distracted.

And so on. Ask any teacher and they can add several more bullet points to this list.

The crucial postscript: Tell students how and why you used AI

I got students seated the next day. (I told each student their new desk number at the door as they came in. I referenced an alphabetical list of names with desk numbers next to them on a Google Doc on my phone — I’ve lost too many physical seating charts to trust myself at this point.)

We spent most of the period doing get-to-know-you activities in their groups to start building rapport. Some were fun (create a group name) and some were more substantial (share memorable life events and how those connect to your career interest today).

Then, towards the end of the period, I said this:

Before class is over, I need to tell you something: I used AI tools to help make these groups. And I’m telling you for two reasons.

First, I want you to know how and why your groups are the way they are. Here are some of the steps I took to leverage AI to make the groups, starting with the survey you completed yesterday…(I then projected and scrolled through my own GPT conversation to show them some of what I shared above.)

I’m telling you all that so you can know how I’m using AI. We need to start having these conversations because no matter what career you explore this year, AI is something you will need to know about. And I’ll teach you how to use AI tools the way I did, especially when we start work on our long-term “purpose projects” later this year.

But the second reason I’m telling you this is you need to know when to stop using AI and instead trust your judgement.

You’re not sitting in the exact groups the AI created. It gave me ideas to start with, but then I turned it off and made the final groups myself. I need to trust my instincts, and you deserve to have groups finalized by your teacher, not a chatbot.

And that brings me to the final point here: I needed to own my work. If your group isn’t working as well as it could, that’s my responsibility to figure out, not an AI tool’s.

We all need to own our work in the AI era. The final product belongs to us, and we build our professional and personal reputations one final product at a time.

---

Later this year I’ll show students how to prompt and how to leverage AI tools in service of their goals, to do work that wasn’t possible (or least not practical) in a high school classroom even just a year ago.

But I’ll also keep reminding them that the most important work we do with AI is the work we do afterwards, without it.

It’s more important than ever for students to leave our classrooms feeling confident in themselves because they know who they are and feeling pride in their work because they know it was theirs.

That’s how we can help ensure our students get a seat at the table in the AI era — or better yet, welcome others to their table and show them where to sit.