Session 01 · All class materials →
Lesson 1 — Quality, Evidence, and Responsible AI
Duration: 120 minutes
Included delivery track: Follow LAB_SETUP.md for the supplied files and local simulation. Required activities use these materials; connected deployments are optional extensions. Instructor debrief: answer key. Real OAuth, cloud releases and payments are simulated or discussed through evidence packets.
Audience: first-time and experienced learners together
Deck: 01-quality-and-evidence
Lab: The AI Claim Detective
Learners will be able to
- distinguish a request, observation, hypothesis, finding, and verdict;
- write a defect report another person can reproduce;
- use an AI assistant without treating its answer as proof; and
- name data/actions that are off limits in a training environment; and
- pitch a right-sized product and distinguish course completion from evidence-based AIQAA certification.
Instructor preparation
Use the Session 01 teaching guide for the opening, staged evidence demonstration, safety wording and recording checklist. Pre-open all linked materials before class; keep the existing two-hour agenda and hands-on time.
Use fixtures/claim-packet.md and the evidence-packet template. E01–E05 support three claim classifications. Capture an analogous local screenshot separately; no database record or console export is supplied. Prepare a few fact/inference/question statements. Read Product Studio, 20 ideas and the proposal template. Show a simple polished fictional app concept, an oversized proposal and how to cut scope. Arrange proposal review appointments before Session 02; don't require all individual pitches during class.
Agenda
| Time | Facilitation |
|---|---|
| 0–10 | Arrival: a beautiful but wrong AI answer. Would you ship it? Explain why. |
| 10–25 | Observation → hypothesis → finding → recommendation. Classify a few statements; confidence is not evidence. |
| 25–35 | Demo a sourced bug summary and identify unsupported claims. |
| 35–50 | Introduce Product Studio and both awards: completion for finished learning activities; AIQAA certification for product quality, development history and GitHub skills. Own repo, weekly reviews and final presentation to Jared; no separate exam. Explore the idea bank and modest scope. |
| 50–60 | Break. |
| 60–95 | Run the evidence lab in pairs; switch driver/verifier halfway. |
| 95–105 | Peer review: identify one claim missing proof and improve it. |
| 105–115 | Sketch a main product journey and name three core features. Explain the proposal/approval handoff before Session 02. |
| 115–120 | Ungraded reflection: one evidence rule, one safety boundary and next step for the pitch. |
Teaching notes
In the existing Product Studio discussion, show the rubric and one cross-project example: the skills standard stays fixed; the expected product behavior is agreed early. The project is the certification assessment, with no separate test. Learners need the roadmap today, not memorization of 21 criteria.
Keep the first session human-centered: good QA protects a customer from a bad experience and protects a team from bad decisions. Do not introduce prompt tricks as the core skill. The core skill is calibrated judgment.
If students finish early, ask them to improve an overconfident AI answer using only citations to the supplied sources. If students need support, give them the sentence starters: “I observed…”, “I expected… because…”, and “This evidence does not establish…”.
Formative coaching
A learner is ready to continue when their packet includes an expected outcome, observed outcome, reproducible steps, and a clear limit on what the evidence does not prove.
Product Studio connection — Pitch your product
Use up to five minutes of the existing closing/debrief time for a project next-step check; do not extend the 120-minute session. The build work happens as homework. Complete the proposal and one rough wireframe after class. Pick from the 20 ideas or propose your own; present the user, main journey, three core features, exclusions and budget to the instructor before Session 02.
Follow the milestone map and review checklist. Record support needs privately; practice lab quality criteria are coaching, not an exam or numeric completion gate.