🤖 What you told me about AI & data analysis

💼 My AI workshop in the Indiana/Illinois area

I’m putting on an in-person workshop just down the street from my school. If you’re in the area, I’d love to see you — and/or some of your teachers!

WORKSHOP: Student Writing in the AI Age

DESCRIPTION: When ChatGPT can do the writing, how do we keep students thinking? Get plenty of real talk and practical strategies here. LUNCH INCLUDED!

📅 DATE: Monday, November 17, 2025
🕣 TIME: 8:30am to 3:30pm
📍 PLACE: The Inkwell, 114 N Jefferson St, Rockville, IN 47872 (about an hour west of Indianapolis)
💵 COST: $99 (includes catered lunch on site)

Want to learn more? More details and registration here

If you’re in Indiana or Illinois and want to attend: You could travel in the morning of the workshop or find lodging nearby.

If you’re farther away but still want to join: You’re welcome to make travel plans!

If you’d prefer a virtual option: I’m planning to turn this workshop into an asynchronous online course in 2026. Stay tuned!

PS: Hit reply if you have questions or want to ask me about it!

📚 New AI resources this week

1️⃣ Here’s How Teachers Really Feel About the Rise of AI in K-12 Education (via Education Week) — Not many schools have policies to guide AI use, and educators remain divided on whether AI should be used in the classroom at all.

2️⃣ The rise of AI tools forces schools to reconsider what counts as cheating (via Associated Press) — “We have to ask ourselves, what is cheating?” says a California Teacher of the Year. “Because I think the lines are getting blurred.”

3️⃣ FTC orders AI companies to hand over info about chatbots’ impact on kids (via The Verge) — Recent reports discuss teens engaging with AI companions shortly before dying by suicide.

📢 Your voice: AI for data analysis

Last week’s poll: How can school leaders use AI for data analysis?

🟩🟩🟩🟩🟩🟩 Making sense of testing data (13)
🟨🟨⬜️⬜️⬜️⬜️ Gathering insights from state reports (5)
🟨🟨⬜️⬜️⬜️⬜️ Analyze attendance and behavior reports (5)
⬜️⬜️⬜️⬜️⬜️⬜️ Strategize class schedules (1)
🟨⬜️⬜️⬜️⬜️⬜️ Other ... (3)

Making sense of testing data: Even though making sense of testing data comes up as one of the first tasks for AI, other uses are much needed too. In the case of taking decisions based on data from reports (attendance, behaviour or state) will be the cornerstone of the new leadership. Time to use data to support smarter decisions.

Making sense of testing data: While I don't like state-level standardized testing, making sense of that data, or classroom assessment data seems to be the most relevant and insightful way to use AI. It's hard to go wrong in using AI in any of these options. — Chad Sussex

Analyze attendance and behavior reports: This maybe crazy but you know me... but something along the lines of using the AI tools to better organize data and files to reduce redundancy. — John Moran

What would you like to read in AI for Admins?

What’s a topic you’d like to see covered here? Hit REPLY to this email and let me know.

Have you done anything you’d like to share with the AI for Admins community? Hit REPLY and let me know.

Would you like to write a guest post to support and equip AI for Admins readers? Hit REPLY and let me know.

🗳 Poll: AI policy status

It’s time to check in. Where is your school/district relating to AI policy?

Instructions:

1. Please vote on this week’s poll. It just takes a click!

2. Optional: Explain your vote / provide context / add details in a comment afterward.

3. Optional: Include your name in your comment so I can credit you if I use your response. (I’ll try to pull names from email addresses. If you don’t want me to do that, please say so.)

📊 AI for data analysis for school leaders

School leaders have more data than ever before.

Test scores. Attendance reports. Behavior logs. Surveys. Budgets.

The list goes on and on …

It’s so easy to get bogged down by all of it. We can literally lose an entire day (or week!) of our lives parsing through and analyzing data. It’s an interesting duality — it can be really helpful OR it can be a time suck.

I recently polled readers like you in this AI for Admins newsletter. A. McLeod put it this way:

“It’s easy to get overwhelmed by the massive amount of data. Having AI put together a quick analysis to get the ball moving is huge. But we still need to be able to look at the data ourselves to be sure nothing’s missing.”

Boom. The human element. That balance. That’s key. AI is the starting point. Human judgment provides the context and creates the necessary action.

Today, I wanted to share some of the ways we can use AI to analyze data to provide insights, save time, and get

The big one: Testing data

Everyone’s most interested in this one — all sorts of stakeholders — and understandably so. State testing data drives decisions, policy, money. It’s high stakes, and not just for students.

And, of course, helpful achievement data doesn’t just come from the state once a year. We have all sorts of touch points … all the way down to formative assessment data from classroom apps that teachers use.

When you start to dump that data into an AI tool like ChatGPT, Google Gemini or NotebookLM, it can really show you some interesting insights.

• Ask AI for insights that you might not have noticed

• Tell it about a focus area and ask for data that relates to that area

• Add data from multiple years and ask for trends

• Add data from other districts, statewide data, or national data to do comparisons — and highlight areas for growth

… and that’s just scratching the surface.

Several of you said the same thing: state testing data is where AI could really help.

The data privacy issue

Technology integrator Crystal Blais touched on the good AND the bad with all of this:

“For so long, I have wanted to use AI tools to assist with testing data! I want a tool to suggest students for interventions or special ed referrals, but I also want to analyze attendance and behavior data, too. My first worry, though, is student privacy.”

That worry is a REAL one. There are a few things to be aware of:

• Some AI models use your data to train the model. If you use the free version (or some consumer paid versions) of LLMs like ChatGPT, your responses can train their AI models — current or future. They usually say something like “used to improve our services,” “retained for training,” etc. if they’re using your interactions to train their AI models.

• Data you share could be vulnerable to a data breach. Think of all of the data these tech companies have … and how powerful data is … and how many people would love to get their hands on it. Imagine if there’s a data breach — and if bad actors got their hands on data we shared with an AI model.

• School/district accounts to AI tools MIGHT not train the model, but still be careful. If you have an enterprise (business) or education account for a major LLM like Gemini or Microsoft Copilot, training is usually disabled by default. Even if that’s the case, you still need to protect student data.

So, that last part … “you still need to protect student data” …

How do we do that? Here are a few ways …

• Anonymize data. Remove names, student ID numbers, and any PII (personally identifiable information) before using it. If you need to connect it back to individual students, maybe replace student names/ID’s with a simple numbering system (i.e. numbering all students 1-500 instead of using their names/ID’s).

• Check for compliance. If you’re not sure, reach out to a rep at the company that provides the app. Ask about compliance to relevant laws (FERPA, COPPA, CIPA, etc.) and even state-level laws.

• Review the settings and options. It never hurts to scroll and scan these to see if something is enabled or mentioned that you didn’t expect — or that makes you nervous. Settings can be a window into how an app works.

Be careful of predictive AI analysis

Something else that can do — that people will ask it to do — is to make predictions. “Based on this data, what is likely to happen?”

The allure of this is strong. It’d be nice to know what’s going to happen — especially if data can clue us in ahead of time.

Here’s the problem. AI models have to make some judgments and assumptions in the background … and sometimes, it’s making a LOT of those assumptions — even on things that we don’t realize it’s considering.

The more assumptions it makes, the more likely it is to veer off course.

Plus, some of those assumptions can be made on biased — and harmful — judgments that it’s made from flawed data.

An example: Imagine that school leaders want to put students on academic tracks based on their achievement (and other) data. When AI starts to make those judgment calls for us, it might take certain things into account …

• their addresses (influencing decisions based on where they live)

• their attendance (which might miss out on context about transportation or caregiver issues)

• their native language (which could skew accurate analysis of students’ cognitive abilities)

When AI is making predictions on student achievement — or who has the best chance to succeed at the school for admissions decisions — those AI predictions can be inaccurate and harmful — and can perpetuate inequities.

The siren’s song of AI predictions is alluring. “Save some time and the struggle. Let AI do it for you.” It can be dangerous.

AI can start conversations (without replacing them)

Data doesn’t have to inform or make decisions. It can give us a starting point for discussions we need to have.

Mike Tornetto (who suggested this topic for the newsletter!) shared a real example of using AI with public data:

“I uploaded [testing data] into NotebookLM and asked it: ‘What overall trends in EOC scores are evident across different districts and years?’ It gave me a great response. Not perfect, but quick, accurate and better than what many people could do on the fly. Even cited the specific data it used to draw each of its conclusions.

“Often school leaders just hand off raw data and tell staff or teachers to interpret or draw conclusions and it flops. People don't know where to begin, start pointing fingers, etc. Sometimes there is no one facilitating/guiding the discussion. To be blunt, people — even data experts — often suck with data analysis, only drawing the most simplistic of conclusions (good, bad, ugh) or not having any more context to dive deeper.

“Would be better practice to give teachers testing data and then show them how to upload it to AI (NotebookLM) and then how to use AI/prompt AI with probing questions.”

Brilliant. Totally agree with this, Mike.

Other ways AI can support with data analysis

While testing was the hot topic, readers also identified other areas where AI could save time:

Attendance & behavior reports: Find patterns (like high absenteeism on Mondays/Fridays). You could also correlate attendance to performance to inform interventions.

Continuous improvement cycles: One reader wrote: “In our province (up in Canada), we have to show engagement with a Continuous Improvement Plan cycle, based on a foundation of responding to data that we have gathered from our student, parent, teacher, and leader stakeholders. We plan to use AI to help analyze data and create an actionable plan in response to trends to inform next steps in our improvement plan cycle.”

Scheduling: There’s so much strategy and arranging that AI can help with here! Class schedules. Room assignments. Bus routes. Two readers said AI for master schedule building would be a game changer—saving days of work compared to clunky SIS tools. Erin Asamoto wrote: “AI for master schedule building. This takes hours and days to get right just to have to redo it later on. Having AI take the number of students, number of classes and class sizes, and specifics in schedule would save days worth of work. Our SIS systems never get this right and adding AI could be a game changer!”

The goal is still maximizing humanity

In all of these cases, it isn’t about AI replacing leaders. We’re using AI to empower leaders to make better human decisions. Making sense of testing data. Tackling scheduling headaches. AI can quickly identify patterns in all of this so leaders can focus on decisions, context, and people.

And that’s the real win: less time buried in spreadsheets, more time supporting students and staff.