Skip to content

Lesson 1 of 10 in AI in education

Where AI pays first in a district or university, and the bar it has to clear

Pick your first AI pilot on the adult side, where a wrong draft costs little, and run the boundary test on grades and student records.

3-minute read, 20-minute exercise. Updated . Free lesson, no account needed.

You leave with: First AI pilot in education (blank card, worked example), three AI prompts.Blank card (PDF)The card as text, with a worked example

In 60 seconds

  • Start on the adult side. Teacher prep, paperwork and back-office questions pay first; a wrong draft costs little.
  • Keep it in a walled garden. The institution owns the data, and no model trains on it.
  • Run the boundary test. Output touching a grade, a student's record, a decision about a student or a conversation with a child or family waits for a named owner.
  • Count the AI you already run. Adaptive-practice engines, plagiarism detectors, early-warning models and scheduling optimizers may already be running without the name.
  • Name the number before the pilot. Measure the same number after, against a bar set in advance.

Kris Hagel, Chief Information Officer at Peninsula School District, counts roughly 30 hours of instructional time back per high school teacher a year. It comes from a voice command that takes attendance. That is where AI pays first: repetitive adult work. Pick your first pilot there.

A free lesson from AI in education, one of its 10 lessons. Read another free lesson

Where the pay actually sits

Where What pays Goes deep in
Back-office and paperwork Contracts, handbooks, handover notes Where AI pays first in government (public sector)
Teacher planning and grading Draft plans a teacher edits Where AI goes into the teacher's week
Tutoring and differentiation Student practice; the boundary bar comes first Where AI gives one-to-one attention
Family communication Due dates and documents Launch a chatbot on one question (Launch the chatbot on one question, because you answer for every reply) and the public-sector front door
Advising and retention Early alerts Put AI into every program
Admissions One college folded it into advising: advisors from application to graduation When every student arrives with an agent
Research The same rule: adult work first What faculty may use AI for in research

The teacher, counselor and advisor shortage has a lesson of its own: Work that can't be staffed (Spot the work that can't be staffed, and size the case).

Let the backlog and error cost pick the first workflow

Peter Olson-Skog, Chief Administrative Officer at School District 197, says teachers will not have more time tomorrow than today. The first pilot is the work piled deepest, where a wrong draft costs little. Back-office paperwork paid early for Brian Sean Graham, Superintendent at Grand Island Central School District. The previous treasurer's four binders of succession material became a notebook a first-time successor could query. Start where your own binders are.

Run the first tools in a garden you own

Run every pilot in a walled garden the institution owns. The data stays yours, and no model trains on it. Gerardo Martinez, Executive Director of Educational Technology at Westminster School District, says teachers learn AI first, not students. Student access comes later, and only through walled-garden tools. Shadow AI is real. At Peninsula, staff pasted student IEP information into public ChatGPT, and the federally required content filters caught it. Give staff a sanctioned studio before they find their own.

Run the boundary test on every output

Ask of every output: does it enter a grade, a student's record, a decision about a student, or a conversation with a child or family? If it does, the named owner and the deeper lesson's bar come first.

Fulton County Schools' two-year AI task force logged over 300 possible use cases, sorted into Explore Now, Explore Later and Explore Never. The never bucket holds anything that would replace the teacher. Run your own list through the same sort.

Courtney Lewis, Director of Enrollment Services at Trinity Valley Community College, puts AI on the repetitive questions, like due dates and where a document goes. Her people keep the worried calls: "You can't combat fear with AI." Keep that split: routine questions to AI, worried calls to your people.

Count what you run, test what they claim

Count what already runs that nobody calls AI: adaptive-practice engines, plagiarism detectors, early-warning models, scheduling optimizers. Buy nothing new until that estate is counted. Hagel says vendor products "generalize to the middle" of the market. In-house tools match what Peninsula believes about instruction for its 600 teachers. Test every claim of classroom-ready AI on your own work first: your hardest unit, your messiest handbook. Count the AI you already have (Count the AI you already have) and Vendor-real AI or relabel (Is the vendor's AI real, or a relabel?, Test an AI vendor on your own work before you sign) go deep.

Measure from day one, book the clock

Name the number before the pilot: attendance minutes, prep hours, subscription dollars. Measure the same number after, against a bar set in advance. When districts ask to copy Peninsula, Hagel asks first: "where does your data live?" He told one district it was a year to two years of getting its data house in order away from seeing any of these benefits. Book the public-district clock before anyone promises a date. Baseline and bar before pilot (Set the baseline and the bar before the pilot starts) and the public-sector lesson Where AI pays first in government go deep.

Exercise (20 minutes): pick your first workflow and fill in the card

Pick one real AI workflow your institution might run first, from your own pile of adult-side work. Fill in what you know; you won't know every line yourself, and the card is how you find that out. Every line you can't fill gets an owner and a date before you close the card. The card download includes a filled-in example of a first pilot that isn't ready yet, and why.

First AI pilot in education (PDF)

A blank card and a filled-in example, one page each.

Open the card (PDF)
Show the three prompts: fill the card, test it, turn it into a plan

Use AI to fill in and test your card

These three prompts go into your institution's approved AI assistant. Use the first to fill in the card one row at a time; it won't guess a number for you. Use the second when the card is done, to hear the case against starting now. The third asks about your institution and your workflow and turns the lesson into a 30-day plan, with a checklist, rubric or brief for your board if you want one. Use placeholders for institution, system and people names; the help doesn't depend on them.

Prompt 1: Fill in my card with me

Paste this, then add your institution and the workflow you have in mind in one or two sentences, and the card below.

Prompt
Before you start: use your company's approved AI assistant, not a personal account. Replace company, system, supplier and people names with placeholders (District A, System X) and round any figures. The help you get does not depend on them.

You are helping me fill in a card that picks the first AI workflow in my institution. The card is pasted at the end.

Rules:
- Ask me about one row at a time, in order, and wait for my answer before moving on.
- Write down only what I tell you. Never suggest, estimate or invent a number, a name, a yes/no answer or a decision for me, even if I ask you to guess. If I ask, say that this row needs a real answer and ask who could give it.
- If I don't know an answer, write "Not known" and ask who should find out and by when. Put that in the Owner / by when column.
- After the last row, show the whole card as a table with three columns: Question, Answer, Owner / by when. Then list every row marked "Not known" with its owner, and remind me that any unanswered row means the workflow is not ready to start.

My institution and the workflow, in one or two sentences: [DESCRIBE YOUR INSTITUTION AND WORKFLOW]

The card:
[PASTE THE CARD]

Prompt from The AI Decision Course by Dr. Tony Hoang, host of The Artificial Intelligence Podcast (Spotify, Apple Podcasts): tonyphoang.com/learn

Prompt 2: Stress-test my card

Paste this, then your filled card.

Prompt
Before you start: use your company's approved AI assistant, not a personal account. Replace company, system, supplier and people names with placeholders (District A, System X) and round any figures. The review does not depend on them.

Below is a card I filled in for the first AI workflow in my institution. Act as a skeptical reviewer at the decision meeting and argue the case for EXPLORE NEVER.

For each row:
- Say whether the answer is specific enough to act on (a number, a named role, a date) or too vague.
- Point out where answers contradict each other, for example what the boundary test says against who owns family-facing or student-facing output.
- Say what evidence you would ask for before accepting it.

Do not rewrite my answers and do not invent facts about my institution. If a row is blank or says "Not known", treat it as not ready.

Finish with: your recommendation (EXPLORE NOW, EXPLORE LATER or EXPLORE NEVER), the three rows that most need work, and one question I should ask my team on Monday.

My card:
[PASTE YOUR FILLED CARD]

Prompt from The AI Decision Course by Dr. Tony Hoang, host of The Artificial Intelligence Podcast (Spotify, Apple Podcasts): tonyphoang.com/learn

To paste the card into prompt 1, copy its rows from the PDF.

Prompt 3: Make it yours

Paste this and answer its four questions.

Prompt
Before you start: use your company's approved AI assistant, not a personal account. Replace company, system, supplier and people names with placeholders (District A, System X, Supplier B).

You are helping a leader apply one lesson to their own situation. The lesson's five decisions are below. Use only them and what I tell you. Never invent facts, numbers, vendors, laws or results about my organization or my sector. Where a number is needed and I have not given one, say what to measure and who should measure it, instead of guessing.

First, ask me these four questions, one at a time, and wait for each answer:
1. My part of education (a district, a university, a school network) and roughly how many people it serves and employs.
2. The AI workflow I have in mind, in one or two sentences.
3. Where it stands today: an idea, a pilot running, or about to launch.
4. What worries me most about it.

Then give me the following. Keep it short and practical:
A. My plan. For each of the five decisions: one action written for my workflow, the role that should own it (a role, not a name), and what should be done within 30 days. If my part of education makes a decision harder, for example student privacy rules, family-facing output or union agreements, say which decision and why in one sentence.
B. Three questions to ask my team on Monday, specific to my workflow.
C. Ask me which one I want: a checklist for my workflow, a rubric to score outputs against the boundary test, or a one-page brief for my board. Then write it, and finish with a prompt I can paste later to update it.
Mark anything that rests on an assumption with "(check)".

The five decisions:
1. Start on the adult side: teacher prep, paperwork and back-office questions pay first; a wrong draft costs little.
2. Keep it in a walled garden: the institution owns the data, and no model trains on it.
3. Run the boundary test: output touching a grade, a student's record, a decision about a student or a conversation with a child or family waits for a named owner.
4. Count the AI you already run: adaptive-practice engines, plagiarism detectors, early-warning models and scheduling optimizers may already be running without the name.
5. Name the number before the pilot: measure the same number after, against a bar set in advance.

Prompt from The AI Decision Course by Dr. Tony Hoang, host of The Artificial Intelligence Podcast (Spotify, Apple Podcasts): tonyphoang.com/learn

Your company's assistant may answer differently, so read what it writes before you use it.

Check yourself

An attendance voice command for teachers, or a tutor for struggling readers: which pilots first?

Show the answer

The attendance command. The tutor speaks to children, so it waits for the boundary bar and a named owner.

A teacher wants to paste student IEP documents into public ChatGPT. What do you say?

Show the answer

Use our walled studio: the data stays ours and no model trains on it.

A vendor demos classroom-ready AI. What happens before any contract?

Show the answer

It runs on our curriculum and our hardest unit, and we read what comes back.

Sources

Quiz

Join to take the quiz and see your score.

  1. Essay grading would save your high school the most hours. How do you pick the first AI workflow instead?

  2. A proposal would have AI draft the letters deciding which students enter a selective program. What do you ask for first?

  3. A request to buy new adaptive-practice licenses reaches your desk this week. What do you do first?

  4. A board member wants the AI savings a neighboring district reports, copied here by spring. What do you ask first?

This is one lesson from The AI Decision Course.

$199 a year (founding price)

Blank card (PDF)The card as text, with a worked example

Membership adds every lesson and its card, the chat on your own case, your notes and progress saved, and a certificate per course.

Try it for 14 days. If it isn't useful, the full $199 comes back.

Paying with a company card? Ask your manager. Your receipt comes from Link, the checkout run by Stripe, and names the membership.

Made by Dr. Tony Hoang, host of The Artificial Intelligence Podcast.

Ask about this lesson

Where AI pays first in a district or university, and the bar it has to clear

    Free preview: 3 questions a day, answered from this lesson. Follow-ups remember your earlier questions on this visit. Don't paste confidential details.

    Ask about this lesson