Curriculum overview
Audience
This curriculum is for staff of Canadian humanitarian and development organizations. Participants may work in program, operations, technical, or leadership roles. Some will be new to AI. Others will already use approved tools at work.
No paid AI account is required. No real beneficiary, staff, donor, or partner records are used in exercises.
Prerequisites
- Basic familiarity with your organization’s data and communications policies
- Ability to join a 90-minute online or in-person session
- Optional: read the one-page AI use decision guide before Session 1
Learning pathway
| Session | Level | Focus | Duration |
|---|---|---|---|
| Session 1 | Level 1 | Responsible AI foundations | 90 minutes |
| Session 2 | Level 2 | Verified work by role | 90 minutes |
| Session 3 | Level 3 | Governance and adoption | 90 minutes |
| Follow-up clinic | All levels | Workplace application | 60 minutes |
Sessions are designed to be delivered once per month. Optional ten-minute practice between sessions uses only fictional or approved low-risk tasks.
Learning outcomes
By the end of the series, participants will be able to:
- Screen a task using a seven-step risk screen and record a green, amber, or red decision with a named owner.
- Compare AI-supported drafts to authorized sources and mark unsupported claims, contradictions, and harmful omissions.
- Improve a prompt by naming the task, audience, permitted sources, output format, and a rule for missing facts.
- Protect data by refusing to enter personal or sensitive operational information into unapproved tools.
- Name human control points including who approves, can override, and provides remedy.
- Draft a bounded pilot charter with a non-AI baseline, consultation requirements, success measures, and a safe exit.
Time plan
| Week | Activity | Participant time |
|---|---|---|
| Before Session 1 | Pre-assessment (optional) | 10 minutes |
| Month 1 | Session 1 + worksheet | 90 minutes |
| Between 1 and 2 | Optional practice screen | 10 minutes |
| Month 2 | Session 2 + worksheet | 90 minutes |
| Between 2 and 3 | Optional prompt improvement | 10 minutes |
| Month 3 | Session 3 + worksheet | 90 minutes |
| Month 4 | Follow-up clinic | 60 minutes |
| End | Post-assessment (optional) | 10 minutes |
Canadian legal and accessibility context
This kit aligns with federal automated decision-making guidance and OPC generative AI principles. Organizations must also apply PIPEDA (and Quebec Law 25 where relevant), provincial privacy laws, and accessibility standards such as the Accessible Canada Act and AODA (Ontario) to their own materials and platforms. References point to primary sources; this kit is not legal advice.
Indigenous data sovereignty and partner-led governance are named in Session 3. Classroom exercises are not a substitute for nation-specific or community-led consultation.
Adapting for 15 to 40 participants
For skills assessment during case practice, facilitators should aim for no more than 8 participants per breakout group so each person can complete the worksheet. At 31 to 40 attendance, use parallel breakouts with co-facilitators or accept worksheet submission within 48 hours for rubric scoring.
Group sizing
| Attendance | Suggested groups | Notes |
|---|---|---|
| 15 to 20 | 3 to 4 groups of 4 to 5 | Start in pairs within groups |
| 21 to 30 | 4 to 5 groups of 5 to 6 | Rotate reader, recorder, risk challenger, source checker |
| 31 to 40 | 5 groups of 6 to 8 | Use pairs for breakout tasks; plenary debrief; consider co-facilitator |
Online delivery
- Share worksheets as editable files or no-login web forms before the session
- Use typed chat as an alternative to spoken breakout reports
- Enable captions; confirm who manages platform accessibility settings at inception
- Do not record breakouts or use AI meeting assistants on participant discussions
In-person delivery
- Provide printed worksheets for participants who prefer paper
- Post slide content on a shared drive at least five business days before each session
- Allow a no-camera equivalent: written participation routes online
Low-bandwidth adaptations
- Distribute case materials as plain text before the session
- Synchronous: phone dial-in plus typed chat; facilitator reads contributions aloud (see technology checklist)
- Asynchronous fallback when live join is impossible: worksheet within 48 hours and debrief window
- Offer a text-only participation route without breakout rooms
- See facilitation tips in each session folder for more detail
Materials included
Each session folder contains:
- Facilitator run sheet (minute by minute)
- Learning objectives
- Slide deck (markdown source, HTML, PDF)
- Speaker notes
- Facilitation tips
- Participant worksheet
- Answer key and debrief questions
Additional resources: eight synthetic case studies, pre/post assessment, skills rubric, evaluation form, decision guide, data-responsibility checklist, facilitator guide, participant handbook, and references.
Assessment approach
Performance scenarios scored with the four-point skills rubric are the primary pre/post measure. Five self-efficacy items are secondary. An optional 12-item knowledge check supplements reporting. See the assessment folder for scoring guidance and how to read small samples constructively.
Delivery and adaptation resources
- Facilitator delivery guide, preparation checklist, and sample invitation email
- Participant handbook and technology and accessibility checklist (captioning plan; synchronous phone-plus-chat low-bandwidth route)
- Visual aids with diagrams and text equivalents
- AAP adaptation design note for organizations adapting the kit (no claim of prior community consultation in this repository)
- Optional Session 2 live-tool lab for org-approved tools with synthetic data only
Evidence matrix
| Requirement | How the kit addresses it | Primary sources |
|---|---|---|
| Responsible AI use | Seven-step screen; G/A/R decisions in every session | IASC 2023; UNESCO 2021; OPC 2023 |
| Data responsibility | Checklist; protection and sharing cases | IASC 2023; OCHA 2025; ICRC Handbook 3rd ed. |
| Verification | Source-check tables; Session 2 prompt lab | Consensus literature on verification gaps |
| Humanitarian protection | Protection lens; chatbot and SMS cases | Fundamental Principles; ICRC AI policy 2024 |
| Governance and adoption | Pilot charter; vendor pitch case | Canada ADM directive; UNHCR AI approach |
| Canadian context | PIPEDA, AODA, ACA references | Government of Canada; OPC |
Licence and use
Content is licensed CC BY 4.0. Code is licensed Apache-2.0. You may adapt, translate, and reuse with attribution. These are learning materials, not a record of delivered training.