Professional Development · A Course for Teachers
After EffortThe disciplined use of AI in teaching
A practical course built on a single idea: a tool helps a student most when it comes after their own effort rather than in place of it. Reached for after a student has already struggled with something, a tool can genuinely extend them. Handed over before that struggle, the same tool quietly removes the occasion on which they would have grown. The course teaches teachers to tell those two situations apart, and gives them a way to act on the difference in an ordinary week of teaching.
01 / The Premise
The danger isn't that AI will fail.
Failure, schools can manage; they have managed it for decades. The real danger is that AI will succeed just enough to conceal what it has removed. It will help students complete the work while sparing them the inconvenience of becoming educated.
Most AI training for teachers answers the wrong question. It asks how to use the tool, meaning which buttons, which prompts, which platform. It rarely asks when, and almost never asks whether. So teachers leave the workshop fluent in a tool and no clearer on the one question that actually protects a child: did the machine arrive before the student had done their own thinking, or after it?
This course takes the opposite order. Steve Jobs argued that the most important thing in education is not a device or a network but another person, a guide who incites, feeds, and directs curiosity. The machine is reactive, and children need something proactive. A chatbot is fluent, patient, tireless, and available at 2:13 in the morning, and it can say "good question." A teacher knows whether it actually was a good question. That difference is not nostalgia. It is most of the job.
Difficulty is not a bug in education. Difficulty is the mechanism. A student who wrestles with a paragraph is not only producing writing; they are discovering what they actually think. AI used badly removes that resistance too early and too easily. Used well, it arrives after the wrestling and makes the next round of wrestling possible. This course gives teachers a shared vocabulary, a diagnostic they can repeat, and a working discipline for keeping the tool on the right side of that line.
It is built by someone who ships AI-amplified software every day and still believes the line is real. The course is not the work of a critic who hates the machine, and it is not a sales pitch for the future by the seat. The aim is to build judgment rather than hand down a catechism, so that teachers can tell a drafting aid from a decision engine, and are trained not only in how to use AI but in when to refuse it.
02 / Learning Outcomes
What teachers will be able to do.
Every outcome is observable, tied to the classroom, and assessed against an artifact the teacher actually builds rather than a quiz score.
Locate the timing line
Given any classroom AI use, decide whether the tool arrives before or after the student's own effort, and redesign it when it arrives first.
Protect the unaided zone
Define, for their own course, the work students must do without a machine, and articulate the reason to students and parents.
Run the diagnostic
Evaluate a proposed AI use through six questions and reach a defensible verdict: admit as servant, redesign, or refuse.
Pilot without chaos
Introduce a tool reversibly and incrementally, with a pre-committed decision to adopt, iterate, or abandon.
Refuse with confidence
Say no to a tool, set transparent disclosure norms, and keep a human in the lead rather than merely in the loop.
Write the policy
Draft a one-page school or department AI Settlement that survives the next vendor pitch and the next budget cycle.
03 / Audience & Prerequisites
For the people who answer for the child.
No coding required. No prior AI experience assumed. If you have stood in front of students and decided what was good enough, you have the prerequisite.
Classroom teachers, K–12 & higher-ed
Any subject. The course works with your real assignments and your real units. You bring them, and we run them through the diagnostic together.
Instructional & curriculum leaders
Department heads, coaches, and coordinators who must set norms others will follow and defend them to families and boards.
School & district leadership
Principals and PD directors deciding policy before the tools decide it for them. The capstone produces a policy you can adopt.
Parents on the educational compact
An optional evening session brings parents back in: reading with their children, asking what they are learning, and resisting the cult of convenience.
Prerequisites: none technical. Participants should bring one or two assignments they currently give and, if available, one AI tool their school is considering or already using. The course supplies everything else.
04 / Formats & Delivery
Scoped to the time you have.
There is one course, offered at four depths. Each format ships at least one classroom-ready artifact per participant, and can be delivered onsite, virtually, or in a hybrid form. Investment is scoped to your context; you can request a quote in the booking section.
The Timing Line
- The premise, the diagnostic, and the closing question
- Live sorting of the room's own AI uses: thief or tutor
- Each teacher leaves with a one-page Timing Map
- Best as the on-ramp before a deeper engagement
Servant or Master
- Modules 1–3 plus a condensed Method
- Hands-on diagnostic clinic with your assignments
- Ship: a redesigned assignment and an unaided-zone definition
- Half-day variant available (Modules 1–2)
The Full Settlement
- All six modules with the capstone Settlement
- Pilot-design lab with timebox-and-eject planning
- Refusal rehearsals and disclosure-norm drafting
- Ship: a signed, one-page school/department AI policy
Ship Every Week
- One module a week, mirroring the SSD rhythm
- Each week ends in a shippable teaching state
- Between-session practice in real classrooms
- Train-the-trainer add-on for district scale
Credit: the course can be structured to award professional development clock hours per your district's requirements; the two-day and cohort formats include the documentation and reflective artifacts most approval bodies require. Accessibility: materials meet plain-language and contrast standards; captioning available for virtual delivery.
05 / The Arc
From principle to policy.
The six modules move deliberately: first the distinction, then the human it depends on, then the diagnostic, then a builder's discipline, then the courage to refuse, and finally the policy that holds it all in place.
The order matters. You cannot write good policy (Module 6) before you can refuse a tool (Module 5), and you cannot refuse credibly before you can diagnose (Module 3), and you cannot diagnose before you can see the timing line (Module 1). None of this is a slogan to memorize. A serious school cannot be governed by adjectives like "responsible," "human-centered," and "safe" pasted over a difficult room like so many verbal scented candles. The course replaces those adjectives with judgment a teacher can repeat under pressure.
| Module | Moves from | Moves to | Ships |
|---|---|---|---|
| 01 | "AI is good or bad" | "AI before vs. after effort" | Timing Map |
| 02 | "The tool can teach" | "The guide teaches; the tool assists" | Unaided-zone definition |
| 03 | "It feels fine" | "Six questions, one verdict" | Three diagnosed uses |
| 04 | "Roll it out and hope" | "Ship small, reverse easily" | Lesson + pilot plan |
| 05 | "Human in the loop" | "Human in the lead" | Disclosure + refusal checklist |
| 06 | "AI as ideology" | "AI as instrument" | One-page Settlement |
06 / The Six Modules
Each session ends with something you can use tomorrow.
Every module names its governing principle, what we do together, and the artifact you ship. No module is theory-only; no module is tool-only.
The Timing Problem
"A tool used after judgment may enlarge the human being. A tool used before judgment may replace the occasion on which the human being would have been enlarged."
Learn the one distinction that determines whether a tool harms or helps, which is when it arrives relative to the student's own effort. Sort a stack of real classroom AI uses into the two cases, before judgment and after judgment. Then work out where difficulty is doing real formative work and where it is only friction worth removing.
A one-page Timing Map of every AI touchpoint in one of your current units, each marked tutor or thief, with a redesign for the thieves.
Guide, Not Assistant
"The teacher is not merely an information source. The teacher is an embodied standard. A chatbot can say 'good question.' Ms. Hill knows whether it was."
Map what only a human guide can do against what the machine can imitate. Defend writing-as-thinking: the bad first draft is not an embarrassment to be skipped but the corpse from which the good sentence is resurrected. Identify each subject's irreducible zone of unaided work.
A defined unaided zone for one course, meaning the work students do without a machine, plus the rationale you can give students and parents without apology.
The Diagnostic
"Civilization is largely the art of making distinctions before catastrophe makes them for us."
Learn the six questions that turn instinct into a judgment you can repeat. Run your own assignments through the live diagnostic (Section 08). Then calibrate as a group on the hard cases, such as the accessibility tool that doubles as a shortcut, or the drafting aid that drifts into a decision engine.
Three of your real AI uses, run through the diagnostic, each with a verdict (servant, redesign, or refuse) and a concrete next step.
Ship Every Lesson: the Method
From Shippable States Development: maintain a teachable state at all times; ship one real thing each session; fix one variable and let the others flex.
Borrow a builder's discipline for trying AI in class without chaos. Walking-skeleton a lesson end-to-end before perfecting any part. Dark-launch a tool with a small group first. Set a timebox with a pre-committed eject. Practice the nightly reflective ritual: did the tool arrive before or after judgment today?
One classroom-ready, AI-aware lesson and a one-page pilot plan with a pre-committed date to adopt, iterate, or abandon.
Refusal & Transparency
"A teacher who cannot say no to a tool is not empowered by it." Responsibility is not an output. It is a burden, and it stays with a person.
Build the judgment to refuse, and the language to refuse gracefully. Write transparent disclosure norms so that hidden assistance cannot corrupt assessment or self-knowledge. Resist the moral laundry, by which institutions let "the system" absorb the blame. Move the human from in the loop, where they are present too late, to in the lead, setting the question, defining the standard, and bearing the result.
A student-facing AI disclosure norm and a teacher's refusal checklist, setting out the conditions under which the answer is simply no.
The Settlement (Capstone)
"The question is not whether AI will enter education. It already has. The question is whether it enters as instrument or ideology."
Turn the course into policy. Draft the six-principle Settlement for your context (Section 11). Confront the equity trap directly: if AI becomes the substitute for talented teachers in the schools that most need them, the future will have perfected the oldest injustice, giving Socrates with Wi-Fi to some children and the FAQ to the rest. Plan student-data stewardship and parental re-engagement.
A one-page school or department AI Settlement, drafted and signed by your team, durable enough to outlast the next product demo.
07 / The Method
Taught the way good software is shipped.
The course practices what it teaches. It borrows its discipline from Shippable States Development, the daily-shipping engineering practice behind this work, and translates it for the classroom. The point is the same in both worlds: forward motion you can reverse, judgment you can repeat, and a deliverable at the end of every session.
Most professional development front-loads inspiration and back-loads usefulness, so teachers leave moved but empty-handed, and the lesson that was "90% done" never reaches students. This course inverts that. As with a codebase kept deployable every day, each session ends in a state you could carry into class without embarrassment. It is not finished, but it works.
| SSD discipline | Translated for teaching |
|---|---|
| Ship every day No exceptions | Leave every session with one classroom-ready artifact. A session without a shippable result is treated as a miss, not a neutral outcome. |
| Constant production parity | Try the tool in real classroom conditions from the start, with your actual assignments and your actual constraints, rather than in a sandbox that hides the friction. |
| The shippable-state invariant | End each lesson in a state where genuine formation occurred, not merely where the task got completed. |
| The ratchet principle | Every admitted AI use must move student capacity forward in some measurable way. It may never move it backward. |
| Feature flags · dark launching | Introduce a tool reversibly, to a small group first. Easy rollback. The class never blocks on an experiment. |
| Timebox with eject | Every pilot has a pre-committed decision date to adopt, iterate, or abandon, decided before you grow attached to the work. |
| The nightly ritual | A short daily reflection: did the tool arrive before or after judgment? What did it strengthen? What did it weaken? |
| The iron law: fix one variable | In any AI-aided lesson you can optimize at most one of coverage, time saved, and depth of formation. Choose deliberately, and protect depth. |
08 / The Live Diagnostic
Six questions, then a verdict.
This is the core instrument of the course, working here exactly as it works in the room. Picture a specific AI use you are considering, such as a chatbot that drafts essays, a tool that solves the problem set, or an app that summarizes the reading. Answer the six questions and read the verdict. This is the muscle teachers build until it becomes instinct.
Evaluate one classroom AI use
Q1Does the tool arrive before or after the student's own effort?
Q2Which capacity does this use mostly strengthen, or weaken?
Q3Is the machine assisting the student's judgment, or replacing the occasion for it?
Q4Does it make the student more capable, or merely make incapacity less visible?
Q5Does this protect a zone of unaided work, or erase one?
Q6Who bears responsibility for the outcome, and is a human still in the lead?
Each answer on the right edge moves the tool toward master. The verdict updates as you go.
09 / The After-Effort Rubric
The diagnostic as a printable standard.
The same six questions, rendered as an assessment instrument teachers and leaders can apply to any tool, assignment, or vendor claim. The left column admits a servant; the right column names a master to refuse or redesign.
| The question | Servant (admit) | Master (refuse or redesign) |
|---|---|---|
| Timing When does it arrive? | After the student's own effort | Before any effort has occurred |
| Capacity What does it do to ability? | Strengthens a capacity the student keeps | Weakens a capacity they still need |
| Judgment Whose decision is it? | Assists the student's judgment | Replaces the occasion for judgment |
| Visibility Capable or just covered? | Makes the student more capable | Makes incapacity less visible |
| Unaided zone What stays theirs? | Protects essential unaided work | Erases the unaided zone entirely |
| Responsibility Who answers for it? | A human remains in the lead | Responsibility dissolves into "the system" |
How to read it: a use that lands in the left column on all six is a tutor, and you can admit it. One or two answers in the right column means you should redesign before use, because the tool is fine and only its placement is wrong. With three or more, the honest move is to refuse it, or to rebuild the task so the machine arrives later. The rubric is not a gate you pass once. It is a habit of making the distinction before circumstances make it for you.
10 / What You Leave With
You leave with a working portfolio.
The full course ships six concrete artifacts per participant, all usable the next school day, and assembled into a portfolio that doubles as evidence for PD credit.
11 / The Settlement
Six principles a serious school can stand on.
The American habit of meeting every new tool with either hysteria or reverence is one of our least attractive traits, and a serious school can afford neither. The capstone produces a settlement instead: a short, durable policy that admits the instrument and refuses the ideology.
Preserve zones of unaided work
Students need to write, read, calculate, and think by themselves, and to discover the exact dimensions of their own ignorance without a machine rushing in to comfort them.
Make AI use transparent
Students disclose when and how a tool was used, not because every use is wicked, but because hidden assistance corrupts assessment and, more importantly, self-knowledge.
Train teachers in when to refuse
Not merely how to use AI, but when to say no. A teacher who cannot refuse a tool is not empowered by it. Refusal is a professional skill, taught and honored.
Treat student data as sacred
A child's work and record are not mulch for venture-backed personalization engines. Stewardship is the default; extraction requires a reason and a name.
Bring parents back into the compact
Reading with children, asking what they are learning, resisting the cult of convenience, and remembering that a child's frustration is not always an emergency requiring technological intervention.
Restore the dignity of teaching
If the teacher is the decisive instrument, behave accordingly. Recruit better, pay better, train better, grant real authority, and demand excellence, without turning the profession into a shrine for mediocrity.
Beneath all six sits one guardrail against the deepest temptation: AI for the poor, humans for the rich. The affluent will keep small classes, human tutors, and AI as one more instrument in a well-stocked orchestra. The danger is that everyone else gets the instrument without the orchestra. A settlement worth signing closes that gap rather than laminating it.
12 / Your Instructor
Someone who uses these tools every day.
A warning like this lands harder coming from someone who loves the machine. As with the figure at the center of these ideas, the credibility comes from knowing the beauty of the tool as well as its limits.
Alex has spent thirty years building software, at NeXT and Apple (working directly with Steve Jobs), and across Disney, Intel, Symantec, Ford, and the NFL. Today he leads manufacturing operations and test at a deep-tech computing startup, where he ships AI-amplified software daily under his own SSD discipline. He is not a critic shouting from the sidelines. He uses these tools, depends on them, and still insists the line between servant and master is real.
He also holds an M.A. in philosophy, with work on public reason and pluralism, and writes publicly on the civic and cultural stakes of technology. This course grows directly out of two of his essays, The Machine Is Not Ms. Hill and The Machine That Arrives Too Early, and out of close, current proximity to real classrooms through his family's teaching life. The result is an unusual pairing: a practitioner who can wire the tool, and a writer who can say precisely why and when not to.
13 / Booking
Are you using the instrument, or has the instrument begun using you?
Bring this question to your faculty. Choose a format, name a date, and tell me about your context, including subject mix, group size, and the tools already in your building. I will scope the engagement and the PD-credit structure to fit, and send you a quote.
A note in the shareware spirit of this work: it represents thirty years of hard lessons, offered without ceremony. The aim has never been to sell a methodology. It is to help the next generation of teachers keep the machine in its place, as a servant admitted carefully, by adults who know where the doors lead.