Appendix G — For instructors
This page is for whoever runs the course next: the author, a colleague, or someone at another school. It holds the parts of the design that students do not need to read: the module catalog and its variants, the session structure, how to run the learning journal, how to ask a guest to speak, and how to adapt the course for another cohort. The design files themselves are in curriculum/ in the repository (private for now; it will be made public).
G.1 The modules
Modules are independent. Each tags the competencies it serves (AIM-1 through AIM-9 and AIM-X, defined in curriculum/competencies.md), and assumes no prior module. That is what makes the course survivable when a third of the class misses week two.
core- in the original design. Minutes are the chapter’s own stated time.
| ID | Chapter | Minutes | Competencies |
|---|---|---|---|
plan-your-learning |
Plan your four weeks with ChatGPT | 60 | AIM-5, AIM-9, AIM-X |
core-foundations |
How these systems work | 60 | AIM-1, AIM-2 |
core-appraisal |
Critical appraisal | 70 | AIM-2, AIM-1, AIM-4, AIM-9 |
core-bias-equity |
Bias and equity | 75 | AIM-3 |
core-patient-communication |
Talking to patients about AI | 60 | AIM-6, AIM-4 |
core-ethics-regulation |
Ethics: what you owe when you sign the note | 60 | AIM-4, AIM-X |
enrichment- in the original design. Minutes are the chapter’s own stated time.
| ID | Chapter | Minutes | Competencies |
|---|---|---|---|
enrichment-agentic-coding |
Agentic coding | 85 | AIM-5, AIM-2, AIM-4 |
enrichment-prompting-lab |
Prompting lab | 60 | AIM-5, AIM-2 |
enrichment-workforce-economics |
Will AI take your job? | 60 | AIM-7, AIM-2, AIM-X |
enrichment-environment |
What AI costs the planet | 60 | AIM-3, AIM-2, AIM-7 |
enrichment-rag-notebooklm |
Grounding AI in real sources | 70 | AIM-1, AIM-2, AIM-9 |
Eleven modules. Each week uses one as its core chapter and offers others as a menu, so the surplus is deliberate. Swap by cohort interest.
G.1.1 The default assembly (fall 2026)
Since October 2026 the chapters are a framework and a menu, not a sequence to teach. Each week has one core chapter and a menu of others. In weeks 2 to 4 the instructor’s two weekly talks are the main event: Monday on the core chapter’s topic, Wednesday on a menu chapter or a topic beyond the book. The core chapter is the companion to Monday’s talk, read after it, or instead of it by a student who missed it. Students spend most of their own time on Learn it with AI.
| Week | Core (Monday talk) | Wednesday talk | Menu |
|---|---|---|---|
| 1 | How these systems work (covered by the live talks) | Week 1 is taught live | No menu; the Thursday hands-on session uses Plan your four weeks with ChatGPT |
| 2 | Prompting lab | Professional work with AI: writing and beyond (a co-director) | Critical appraisal; Bias and equity; Grounding AI in real sources |
| 3 | Talking to patients about AI | To be announced: bias and equity, or building a small tool live | Agentic coding; the scribe open lab |
| 4 | Ethics: what you owe when you sign the note | Where this is going (jobs, the planet) and what you will keep | Will AI take your job?; What AI costs the planet |
Coverage check: the core chapters make AIM-2, AIM-4, AIM-5, and AIM-6 main subjects. AIM-1 is covered by the week 1 live talks. AIM-3 (bias) is not in a core chapter; it is covered by the bias chapter on the menu and by the week 3 Wednesday talk if it is chosen. The Friday groups are open discussion, so they cover bias only if students raise it. This is a deliberate trade, recorded in curriculum/design-constraints.md.
G.1.2 Variants
A cohort that wants more evidence and appraisal. Make critical appraisal the week 2 core and the prompting lab a menu item.
G.1.3 Four ideas that recur
Modules are independent, but four ideas recur across them. Name the first three in week one, preview the fourth, which week 4 explains, and call them by name whenever they reappear. They are what turns separate chapters into a course, and they are what students will still have when every tool taught here has been replaced.
- Proxies fail. Almost every AI failure in medicine is a stand-in that looked reasonable. Cost stood in for need in the Obermeyer algorithm. Light skin stood in for normal reflectance in pulse oximetry. Race stood in for unmeasured biology in eGFR. A benchmark score stood in for clinical performance in FDA clearance data. The question to train is always: what is this actually measuring, and what was it substituted for?
- Efficiency gains get eaten by volume. Per-unit improvement is cancelled out by growth in use, and this appears in three separate literatures. Google cut per-prompt energy while total emissions rose. AI made radiologists more productive and imaging demand grew faster. AI scribes cut documentation time per note, and the time went into seeing more patients. Efficiency is not the same as less.
- The published number is the optimistic one. Before-and-after pilots on self-selected early adopters beat randomized trials. Vendor benchmarks on curated hard cases beat real-world deployment. The gap between the most prominently reported statistic and the trial result is the most reliable teaching material in this field.
- Who benefits, who bears the risk, who decides. Introduced in the week 4 ethics chapter. A scribe relieves the clinician; the errors land in the patient’s record, and the patient is told at the door. A data center’s costs fall on neighbors who were not asked. Displaced transcriptionists have the least money in reserve. The question to train: are these three the same people, and if not, what closes the gap?
G.2 Running a week
Weeks 2 to 4 have the same shape, so students always know what comes next. Week 1 is taught live on its own schedule (see the syllabus).
| When | What |
|---|---|
| Monday, 60 min, live and recorded | Opens with 20 to 30 minutes of “How I used AI this week”: the instructor’s own real, unpolished tasks, talked through on screen, including one that went wrong and how it was caught. End that part with a dare: try one of these this week. No patient data or confidential review material on screen. Then a talk on the week’s core chapter topic, with live demonstrations. |
| Wednesday, 60 min, live and recorded | A talk on a menu chapter or a topic beyond the book, sometimes by a guest. |
| On their own | The core chapter, suggested, after Monday’s talk or instead of it (about 1 hour), and Learn it with AI (2 to 4 hours). Menu chapters and open labs are optional. |
| Through the week | Guest talks. They count toward students’ 3 hours of sessions; there is no separate guest-talk quota. |
| Thursday, 45 min, live | Office hours: students bring what they are stuck on. Not recorded. Learning journal entry due that night. |
| Friday, 60 min | Small group, expected: open discussion, with topics the students bring. Not recorded. |
The instructor’s cost is real: about 6.75 live hours a week (Monday, Wednesday, Thursday, and four Friday slots) plus two new talks to write. An instructor with less time can run the Wednesday talk as an open lab or hand it to a guest.
Running notes:
- Lead with use, not warnings. Students’ feedback in earlier years was that they were afraid to use chatbots. The safety rules are short and stated once; checking is taught as part of using the tool well, not as a reason to avoid it.
- Everything uses synthetic data. No real patient information in any course activity, in any account.
- Pass/fail means every hour has to be worth it. Each live session should leave students having done something, not only having heard something.
G.3 Running the learning journal
The student page has the five questions. Question 5 asks which chapters, talks, and other resources students used and whether they would recommend them, so a quick tally each week shows what is working. Mark each entry complete or incomplete. In weeks 1 and 2, reply with one sentence: what to try next. Use Friday’s small group, not written feedback, for anything longer. Shared resources that are worth keeping go into the starter resources appendix after a quick check.
The week 4 piece (“what you will keep”) is presented in five minutes at the last Friday group. That session is the course’s close. A student who cannot come submits the piece, and that is enough.
G.4 Asking a guest to speak
The original design capped guests: one per session, two across the whole course. Every session paired an idea with an activity, and a guest-heavy course turns back into lectures without anyone deciding to. In fall 2026 the cap was lifted for weeks 2 to 4: guests give 45 to 50 minute talks plus questions, and students choose which talks to attend.
The most useful guest is someone who can talk about a tool actually deployed in the local health system in the last quarter. That is what a fourth-year audience wants and what a survey lecture cannot give. Good matches by module: ambient documentation as deployed (ethics); disclosure conversations and a patient-facing tipsheet (talking to patients); how a tool gets approved and bought locally (appraisal); the device boundary and local policy (ethics); a proxy failure in a model that was really deployed (bias and equity); what AI is doing to one specialty’s job market (workforce).
The invitation should say: the date and time; the length of the talk; the competency the session serves; Zoom only; the number of fourth-year students (49 in fall 2026); pass/fail; and the actual ask, bring one thing you deployed and one thing that went wrong. Sending the invitations as one batch with a dean’s co-signature gets more acceptances. Sending them one at a time does not.
G.5 Adapting the course for another cohort
Everything specific to the original offering (people, dates, local accounts) is kept out of the chapters on purpose, so the book can be reassembled elsewhere. To run it for your own cohort:
- Check the design constraints in
curriculum/design-constraints.mdagainst yours. The chapters assume about forty to fifty fourth-year students, four weeks, Zoom, pass/fail, and free tiers. - Choose an assembly. The default or one of the variants. Update the syllabus tables and the due-date placeholders.
- Confirm tool access. The chapters assume one general-purpose chatbot without a signup step. Several
[VERIFY]flags in the chapters are access details to confirm the week before: an institutional NotebookLM login, a student plan’s status, an episode still being online. The open flags are listed in one GitHub issue on the repository. - Set your own institution’s policy text where a chapter says to check it, for example the attestation language for ambient documentation in ethics, and your state’s disclosure statutes in talking to patients.
- Choose your guests, if any, using the section above.
- Write the exemplar week of the learning journal yourself. Students copy what they see.
Chapters are reviewed against six invented readers before they change; the Review personas appendix has them, and the How this book was made appendix says how the review runs. If you change a chapter, review it as the six.