October 2026 delivery note: at LAMP, map these learning objectives onto DSSC’s existing workshops before scheduling. Do not run a duplicate AI-literacy course. The small-business variant uses public business FAQs or checklists, never customer records or payroll. Confirm funded stipends and accommodations in each session listing. These are facilitator plans, not announced dates.
Each session is 90 minutes. Times are guides, not rules. Materials lists assume 20 participants. Every handout referenced here is in the printable facilitator kit; how to run the room (including hard moments) is in the Facilitator Guide. Full background for facilitators is in the Workshop Playbook and the Level manuals in docs/.
Every session: welcome and food (10 min), name tags with first names only, an optional sign-in sheet, feedback slips at the door, a clear statement that nothing typed or said will be recorded or stored unless someone chooses to share it, and a close that asks "what's one thing you'll do differently this week?"
Learn 1: What AI is, and what it isn't
Goal: Replace magic and fear with an accurate mental model.
| Time | Activity |
|---|---|
| 0:00 | Welcome, food, introductions ("one thing you've heard about AI") |
| 0:10 | The phone keyboard. Everyone opens a text message and taps the middle suggested word 10 times. Read a few out loud. That is the core idea: prediction, not knowledge. |
| 0:25 | Live demo: the confident wrong answer. Ask an AI chat tool about a specific local building or street history. Then check it together against a real source. Find the invented details. |
| 0:45 | What it's good at / bad at. Group sorts cards: "summarize a long document" (good, with checking), "tell me the law" (bad), "draft a letter" (good, with editing), "decide who gets help" (never). |
| 1:05 | The intern rule. Treat AI like a fast, overconfident intern: useful drafts, always checked. |
| 1:15 | Hand out the privacy checklist. Close. |
Materials: projector, one AI chat account on the facilitator's device (never participants' personal accounts), sort cards, privacy checklist, quick check (before).
Learn 2: Scams, privacy, and checking the machine
Goal: Protect people from the harms most likely to reach them this year.
| Time | Activity |
|---|---|
| 0:00 | Welcome, recap Learn 1 in one sentence |
| 0:10 | Voice clone scam. Explain how a few seconds of audio can clone a voice. Everyone writes a family safe word idea on their card (they do not share it). |
| 0:25 | Spot the scam. Printed examples of AI-written emails and texts that use local street and business names. Circle the tells. |
| 0:45 | What never goes into a chatbot. SSN, bank info, passwords, medical details, leases, private documents. Why: free tools may keep and learn from what you type. |
| 1:00 | The Triple-Check Rule. Any fact, number, law, or link from AI gets checked against the original source before you rely on it. Practice on two AI answers. |
| 1:15 | Hand out the pocket cards. Close. |
Learn 3: Public data and your block
Goal: Show that a lot of civic information is public, and teach the difference between public records and private people.
| Time | Activity |
|---|---|
| 0:00 | Welcome |
| 0:10 | What's public? Tour the city's open data portal and 311 map on the projector. |
| 0:25 | Treasure hunt (worksheet). In pairs, find something on or near your block: an open 311 request, a permit, a dangerous building listing, a council agenda item. |
| 0:50 | The line we don't cross. Walk through the Safety Doctrine. Discuss: "Map of every unresolved streetlight complaint" (fine) vs. "list of which neighbors complain most" (never). |
| 1:05 | Invite problems. "Bring one real, stuck problem to the Solve clinic." |
| 1:15 | Quick check (after). Close. |
Solve Clinic 1 and 2
Goal: Turn frustrations into Problem Briefs with a clear next move and an owner.
| Time | Activity |
|---|---|
| 0:00 | Welcome. Clinic 2 only: report back on every Clinic 1 problem, even the ones that didn't move. |
| 0:10 | Pink notes: dump the frustrations. One concrete problem per note. ("Streetlight at Linwood and Prospect out since July," not "the city ignores us.") |
| 0:25 | Yellow notes: who has authority? Landlord, city department, utility, county, state, or community. |
| 0:40 | Green notes: what's the evidence? 311 cases, permits, violations, meeting minutes, photos with dates. |
| 0:55 | Pick 3-5 and write briefs in small groups using the Problem Brief template (below). AI can help draft; a human checks every line. |
| 1:15 | Commitments. Each brief gets a named next step and a date. Post them on the public problem log. Close. |
Problem Brief template (printable, or online); track all of them on the public problem log:
- The problem, in plain words. Where, since when, who is affected.
- Who has authority to fix it (specific office or entity).
- What evidence exists (with links).
- The smallest next move this week, and who will do it.
Never write briefs about individual private residents. Focus on institutions, commercial landlords, and infrastructure. See the Level 2 manual.
Build Session(s)
Goal: Co-design one small tool from the strongest brief, safely.
| Time | Activity |
|---|---|
| 0:00 | Welcome. Read the chosen brief out loud. |
| 0:10 | Rubric check. Score it together with the score sheet (full detail: Pilot Rubric). If it fails the Red-Light Test, stop and pick another. |
| 0:25 | Paper prototype. Draw the screens on paper cards. Test them on each other. |
| 0:55 | Write the rules. Plain-language instructions for any AI piece: always link sources, never give legal advice, say "I don't know" when unsure. |
| 1:10 | Name the human reviewer. Who checks the tool's output, and how often? |
| 1:20 | Close. Technical volunteers take the spec to Building a Tool. |
Good first tools: a service navigator for one problem type, a meeting-commitment tracker, a resource map, an intake checklist. Bad first tools: anything that decides, scores, predicts, or tracks people.