Facilitators

Run the lab as a practical discussion session.

Facilitation should help learners compare reasoning, notice uncertainty, and connect the generic scenarios to local policy and work patterns.

The self-guided path is intentionally sequenced. Facilitator preview opens the modules for planning and workshop navigation without marking learner work complete.

Preview mode is for facilitators, trainers, and curriculum reviewers. It should not be used as the learner completion path.

Open facilitator preview

Facilitator stance

• Ask learners to commit before introducing fuller context.
• Use Before You Act prompts to broaden what they notice.
• Normalize revision or clear defense as evidence of learning.
• Require a documented judgment, not just a verbal preference.
• Keep discussion focused on evidence, data boundaries, accountability, and escalation.
• Redirect legal, HR, privacy, and security questions to local experts.

Readiness checklist

• Name the audience, required outcome, and whether attendance is optional or required.
• Confirm approved AI tools, prohibited uses, and data classification examples.
• Identify privacy, security, legal, HR, compliance, and procurement escalation contacts.
• Decide how completion will be handled before learners start the lab.
• Prepare two or three local examples that are realistic but do not expose sensitive data.
• Test facilitator preview, learner module access, printing, and local storage behavior.
• Plan accommodations for learners who cannot store browser data or submit a record.

Sensitive-scenario warning

Some scenarios ask learners to reason about confidential, regulated, identifying, benefits, personnel, procurement, legal, or safety-sensitive work. Do not ask learners to disclose real client, patient, employee, applicant, student, resident, vendor, case, account, or incident details during discussion.

Use fictionalized or abstracted local examples. When a learner raises a real sensitive situation, pause the group discussion and route the question to the appropriate privacy, security, legal, HR, compliance, program, or clinical owner.

60-minute run sheet

• 0-5: frame AI literacy as practical judgment, not tool expertise.
• 5-10: review local approved tools, prohibited data, and escalation contacts.
• 10-25: discuss Module 1 capability judgments and where confidence was misplaced.
• 25-40: discuss Module 2 verification failures and required evidence checks.
• 40-52: map Module 7 escalation choices to local policy and contacts.
• 52-60: capture one local workflow that needs clearer AI guidance.

90-minute run sheet

• 0-10: frame goals, completion expectations, and local guardrails.
• 10-25: Module 1 discussion on useful, weak, and inappropriate AI tasks.
• 25-40: Module 2 discussion on hallucination, verification, and source quality.
• 40-55: Module 3 discussion on confidential, regulated, and identifying data.
• 55-70: Module 7 discussion on escalation thresholds and accountable owners.
• 70-85: small groups rewrite one local workflow with review and escalation steps.
• 85-90: confirm next steps, open policy questions, and completion handling.

180-minute run sheet

• 0-15: orient learners to the lab, local addendum, and session norms.
• 15-45: Modules 1 and 2 as guided practice with group debriefs.
• 45-55: break.
• 55-95: Module 3 data boundary practice and local data classification mapping.
• 95-125: Module 5 accountability review and human-owner discussion.
• 125-135: break.
• 135-165: Module 7 escalation practice using local contacts and thresholds.
• 165-180: capture policy gaps, workflow changes, and learner completion steps.

Half-day run sheet

• 0-20: orient learners, confirm local addendum, and run a short readiness check.
• 20-70: Modules 1 and 2 with individual commitments, Before You Act prompts, and revision notes.
• 70-80: break.
• 80-135: Modules 3 and 4 with local data, fairness, and harm examples.
• 135-145: break.
• 145-205: Modules 5 and 7 with accountability and escalation mapping.
• 205-235: Module 8 capstone decision memo or team version of the memo.
• 235-240: confirm completion handling, unresolved questions, and follow-up owners.

Module debrief scripts

• Module 1: Name one task where AI was useful, one where it was fragile, and the signal that changed your judgment.
• Module 2: Which confident claim needed verification, and what source or owner would you check before relying on it?
• Module 3: What information should not enter an AI tool, and what safer route would preserve the work goal?
• Module 4: Who could be affected by this output, and what assumption, proxy, or omission needs review?
• Module 5: Who owns the final decision, and what would meaningful human review require before use?
• Module 6: Where can AI reduce friction without removing judgment, documentation, or review?
• Module 7: What threshold tells you to proceed, modify, pause, document, or escalate?
• Module 8: What recommendation can you defend, what evidence supports it, and what remains unresolved?

Misconception bank

• AI is objective. Ask what data, incentives, omissions, and review steps shape the output.
• Confident prose means verified work. Require evidence checks for dates, numbers, names, citations, and commitments.
• Public information is always safe to paste. Separate public facts from confidential context, identifiers, and internal intent.
• A draft cannot cause harm. Track where drafts influence decisions, records, messages, or downstream automation.
• Human review is a quick skim. Name the accountable reviewer and the standard they must apply.
• Escalation slows the work. Treat escalation as routing uncertainty to the right owner before impact grows.
• Lab completion approves tool use. Clarify that approval comes from local policy, not this curriculum.

Local policy addendum template

• Audience and scope: who the guidance covers and which workflows are in scope.
• Approved tools: tools learners may use, account requirements, and any disabled features.
• Prohibited inputs: confidential, regulated, identifying, contractual, security, or personnel data that must not be entered.
• High-risk uses: decisions or recommendations involving rights, access, safety, money, employment, or legal obligations.
• Required review: facts to verify, sources to check, and records to retain before sharing AI-assisted work.
• Escalation contacts: named channels for privacy, security, legal, HR, compliance, procurement, and tool support.
• Completion handling: whether the lab is optional, recommended, or required and how records are submitted.
• Owner and date: responsible policy owner, effective date, and review cadence.

Completion handling

• Treat the public learning record as self-attested practice, not identity-verified certification.
• Do not use facilitator preview as a learner completion path.
• If completion is required, collect the record through an existing learning system or documented internal process.
• Tell learners what to submit, who can see it, how long it is retained, and whether reflections are reviewed.
• Provide an alternate attestation route when local storage, printing, or browser persistence is unavailable.
• Avoid silent collection of learner reflections or browser data from the public version.

Session outputs

• One list of local policy gaps or ambiguous workflows.
• One set of escalation contacts and thresholds that learners can use immediately.
• One completion-handling decision communicated before the session ends.
• One follow-up owner for unresolved policy, privacy, security, HR, or legal questions.