Make the lesson more useful. Keep the learning goal in view.
AI can give you a revised handout to examine. The important question is whether its changes help students do the thinking your lesson is meant to teach.
By Joe McKenna. Adapted from his September 29, 2026 workshop.
Review a revised lesson against its learning goal, source and assessment before using it with students.
A fast revision raises a better question
In the workshop, I tell a story about working with professors who had concerns about AI. One professor told me he had reworked his course materials during the session, after having left them unchanged for several years because revising them was such a large job.
That was his account, as I recalled it in the workshop. It was not an independent evaluation of the revised curriculum, and the recording contains no comparison of student learning before and after. What interested me was the question it opened: how do we judge the work when making a draft becomes easier?
For an educator, a useful starting point is smaller than rebuilding an entire course. Choose one lesson that needs attention. Ask what students should be able to do with it, then use AI to help produce a revision you can evaluate.
A faculty story that raises a learning question
What to notice: This is Joe recounting an earlier experience, not independently measured curriculum quality or student outcomes. Use the story to ask what changed in the learning, not just how quickly material was produced.
1:10. From The One-Person AI Business, 26:13.40–27:23.20.
Visual context: Joe recounts a faculty workshop experience beside the shared browser. No independent faculty testimony or student results appear in this excerpt.
Read the clip transcript
I'll give you a back story. I was back when I was just coming out, and a bunch of professors asked me to come talk about it. I realized that I was actually being invited to be ambushed, and they were going to tell me about how terrible AI was for education and all that. I started by saying, "No kidding, only thing I'm doing this. AI is going to be terrible for education, but it's here and there's nothing to do about it." I convinced them to take one of my classes. The second class I taught them what I'm going to teach you guys right now, and one of them interrupted me 5-10 minutes in the class, interrupt me and said, "Joe, I just rebuilt my entire course curriculum and all of the content while you were talking to me." I says, "Well, what does that change for you in education?" He goes, "Well, my content was four years old. What I was teaching was four years old because I did not have time, spent a whole summer rebuilding it." He says, "And I just built a whole thing and it's better than I ever made." I said, "So did you cheat?" He goes, "What do you mean?" I said, "Well, what if your students did that? What if they made their content? What if they got their answers in five minutes? Did they cheat? It sounds like you cheated at making your content." That whole group of professors in that class changed their entire perspective and now they're very AI forward.
Automatically transcribed from this excerpt and aligned locally. Transcript accuracy and word synchronization still require a human listening review.
Name the thinking you want students to do
A clearer handout can still teach the wrong thing. If your goal is for students to justify a decision, replacing the activity with definitions and a summary may remove the practice they need.
Before requesting changes, write down the intended action: compare two explanations, identify evidence, calculate a result, or defend a choice. Then ask what work would show that action. Carnegie Mellon’s Eberly Center recommends aligning learning objectives, teaching activities and assessments. That is a useful check on an AI revision as well.
Keep the existing requirements visible: your approved sources, the students’ prior knowledge, available time and what you will assess. Ask AI to flag gaps. Do not let a plausible addition quietly become a course requirement.
Try a revision you can inspect line by line
The original activity is open enough that a student could jump straight to hiring more staff. That might be a proposal, but these notes do not establish its effect. I would ask for a revision that makes the evidence work visible.
Revise this fictional activity for a 15-minute introductory class discussion. Preserve the learning goal and every number in the source. Have students separate observations, possible explanations and missing information before proposing a next step. Do not supply their final answers. Label any new scenario details as proposed additions, and explain how each activity step serves the learning goal.
The revision makes the intended reasoning visible without adding café facts. For example, more orders coincided with a longer average wait in these two periods. That observation alone does not establish what caused the difference. Order complexity and interruptions remain unknown; they are not evidence that either caused the wait.
Look for what improved—and what disappeared
Compare the revision with the original. The numbers and staffing count should be unchanged. The new sequence should give students a chance to practice separating evidence from an explanation. It should also fit the available time; five questions might be too much for a group that needs a worked example first.
Use these checks before bringing it into class:
- Accuracy: can every stated fact be traced to the source?
- Alignment: does the task require the thinking named in the goal?
- Challenge: has the revision supplied an answer students were meant to develop?
- Clarity: will learners know what to produce and how it will be judged?
- Access: is the essential information available as readable text, with any diagrams explained?
For this example, an educator’s review guide could look for accurate observations, clearly labeled hypotheses and a proposed check that addresses missing information. A confident recommendation with no evidence would need further work. This is a suggested assessment approach, not a validated rubric or proof that the revision improves learning.
Separate preparing a lesson from assessing a learner
My faculty story leads into a discussion about what counts as doing the work. It does not settle an institution’s AI policy. An educator using assistance to prepare materials and a student being assessed on a particular skill have different responsibilities.
State the boundary for each assignment. If students must independently distinguish evidence from inference, decide whether AI assistance would replace the very work being assessed. If evaluating an AI response is the learning task, a supplied response may be appropriate. Check the applicable course and institutional rules before designing either activity.
Where AI use is permitted, consider asking for the original source, a short account of the assistance used, one accepted or rejected suggestion and the reason. Those items give you something to discuss with the learner. They do not, by themselves, prove authorship or understanding.
Keep real student records and identifiable work out of a practice exercise unless your institution has approved that use and the chosen tool. Fictional examples let you examine the teaching task without needing those records.
Revise one lesson, then decide whether to keep it
Pick one handout you know well. Save the original, name its learning goal and request one bounded revision. Use the targeted-editing workflow if most of it already works.
Read the revision as a teacher and as a student. Identify one change worth keeping and one you would reject or question. If you use it in class, compare the resulting student work with the goal you set; do not infer improvement from faster preparation alone.
You do not have to rebuild everything to find out whether this is useful. A checked change to one activity gives you a concrete decision to make—and a better starting point for the next revision.
Questions that come up
Should I ask AI to rewrite the entire course?
Start with one lesson whose source and learning goal you can inspect. A whole-course draft creates more material to review and can hide changed requirements. Expand only when you have a workable review process.
Does the professor’s story prove that AI improves teaching?
No. It is Joe’s recollection of a professor’s report, not an independently assessed result. The practical question is whether a specific revised activity supports its intended learning, which requires examining the activity and student work.
Can students use the same process for assignments?
Only within the rules for that course and assignment. The educator should specify which assistance is permitted and what students must demonstrate themselves. This article does not create or override an AI-use policy.
15 Days of AI
Put this into practice with Joe.
Give AI useful context. Work with your own materials. Create something you can use. Explore the independent classes and choose the ones that fit your goals.
Explore the classesAbout this lesson
First-person adaptation of Joe McKenna’s automatic workshop transcript, with the faculty account explicitly presented as his recalled anecdote. No direct quotations or claims of audio verification. The café source, prompt, revision and evaluation guidance are authored teaching examples, not demonstrated classroom activities or measured outcomes. The article does not reproduce the transcript’s unverified broad claim that the group’s perspective changed, endorse its rapid-curriculum quality claim, or establish institutional policy.
Source passages: 26:08–27:38; 29:39–31:26; 45:18–46:04. Source: The One-Person AI Business, September 29, 2026.
Official guidance checked September 30, 2026. Product features can change.
AI-assisted organization and supplementary examples. Prepared from Joe McKenna’s workshop.