What to assign, and when
The adoption quickstart: a one-page schedule for instructors.
Students will not go looking for the student guide. AI is free and takes no effort to start using, and ease crowds out intention. A student who has never been asked to think about how to use AI well has no reason to seek out a guide on the subject. You are its only reliable route.
Everything listed below already exists. Nothing here asks you to write anything.
If you assign only three things
students/start-here.md, in week one.companion/spp/, before the first assignment where AI is permitted.students/checklist.md, before they submit that assignment.
Ten minutes, two minutes, three minutes. That is the whole minimum, and it is enough to move a student from fumbling to deliberate.
The full schedule
| When | Give them | Their time | What it does |
|---|---|---|---|
| Week 1 | students/start-here.md |
10 min | The on-ramp. What AI is good and bad for, and the one rule: try to solve. |
| Week 1 | companion/dpp/ |
10 min, once | They configure their AI’s persistent settings. A one-time setup that improves every conversation after it. |
| Before the first AI-permitted assignment | companion/spp/ |
2 min | A paste-in prompt. It turns an answer machine into a study partner that asks what they have already tried. |
| Before the first AI-permitted assignment | Your AI-use category, in writing | none | Say which of the four categories applies. instructors/briefings.md has three ready-to-drop syllabus policies; instructors/assignment-templates.md has per-assignment language. |
| Week 1, if you want a signed agreement | companion/compact/ |
5 min | The Student Compact, in signable form. Some courses use it, some do not. |
| First time they work a real problem with AI | A study session from students/study-sessions/ |
15 min | A worked walkthrough in statistics, programming, physics, or mathematics. Pick the one that matches your course. |
| Mid-term, if they are drifting | students/from-answers-to-learning.md |
10 min | For the student who is extracting answers and not learning. |
| Before they submit | students/checklist.md |
3 min | Ten items. The last is the blank-page test, and it is the strongest one on the list. |
| Before they submit, if you require disclosure | students/disclosure-templates.md |
5 min | Templates. They fill one in and attach it. |
| If your policy says you may ask to see the AI thread | students/sharing-ai-conversations.md |
5 min | How to share a conversation with you. All three syllabus policies in briefings.md reserve this right, so students need to know how. |
For you, not for them
| Read | When |
|---|---|
instructors/first-time-guide.md |
Before the term, if you have not taught with AI before. Start here. |
instructors/briefings.md |
When you write the syllabus. Three ready-to-drop AI policies: restrictive, default, permissive. |
instructors/assignment-templates.md |
When you write an assignment. Reusable patterns that build in verification. |
instructors/rubrics/ |
When you set up grading. |
instructors/assessment-models.md |
When class size is your binding constraint. |
instructors/faculty-faq.md |
When a colleague asks you something you cannot answer. |
instructors/semester-redesign.md |
When you are ready to rebuild a course rather than patch one. |
Make it stick
Students treat an optional reading as optional. Tie each one to something that has a date on it.
- Tie the setup to a deadline. “Bring a configured AI to Friday’s lab.”
- Tie the checklist to the submission. Make it a required attachment or a line on the cover sheet.
- Tie the study session to the first assignment where AI is allowed, not to a week with nothing due.
A reading with no hook is a reading nobody does. That is as true of this page as of any other.
These materials are free and MIT-licensed. Adapt them, translate them, put your course name on them. They accompany Learning with AI: A Framework for Students, Instructors, and Universities (James M. Hyman, SIAM Books).