How to use this lesson
This is an AI × PBL project method lesson using Shiguang Huaxia as an example. It invites learners to start with a person or situation they care about, clarify the need, try once using existing resources and AI, and decide the next step based on real feedback. The app is still in beta, and the project for older adults has not officially started. The case is shared to discuss the process and choices, not packaged as a proven success.
University students are not in the same place, so the focus is on online self-study. study.html is the main entry point, with five sections and 17 reading steps. Each step includes an explanation, a task, and self-check tips. It takes about 60 minutes and can be done in parts. Teachers give advice through agreed online communication; fixed pairing is not required, and teachers are not expected to be online at all times.
If you teach via video conference, you can use classroom.html. Keep the 17-page, 60-minute rhythm of teaching and activities. Allow quiet time for writing, and make sharing voluntary. Lower secondary students can draw cards independently or choose to test each other. If a task requires going out or contacting strangers, ask an adult for help.
This lesson does not include app source code, install packages, real configurations, or private communications. Cases are anonymized and do not name specific individuals. Scenes in images are AI-generated, not actual teaching or trial sites. The AHALab logo uses the original image provided by the user.
What we want learners to take away
The driving question is: How can I turn one thing I care about into a small step that others are willing to try and I am willing to own?
After the lesson, learners should be able to write down a specific person and situation, distinguish between observation, guess, and solution; explain what they want to protect using one trade-off; give AI a task with clear boundaries and acceptance checks; understand where one piece of feedback might point; and plan one action they can start within 48 hours.
The product’s ideas and guiding choices must show up in decisions. Who you care about, what you protect, and what evidence would change your mind. Just writing “warm” or “soulful” does not help decide the next step.
First, understand the current state
What we see now is that there is little discussion or submission. Some people mention difficulties, and some have not yet chosen what they want to do. Reasons might be that the topic is not a good fit, the task is too big, knowledge or resources are lacking, time is tight, or there is pressure from evaluation. It might also be that people are more used to writing before speaking. Ask about specific conditions first. Do not infer personality, curiosity, or mental state from chat activity.
| Teacher concern | Reframe as an observable question | Help for learners |
|---|---|---|
| Noted a problem but did not ask follow-up questions | Can you write one scenario and two questions you do not know the answer to? | Start with a recent experience |
| The problem did not become a need | Can you tell the difference between observation, guess, and function? | Use Action Card A to rewrite the need hypothesis |
| The idea has not been developed yet | Can you write both a long-term vision and a small step for this week? | The vision focuses on value for the people involved; the first step reduces the cost of testing |
| Has not started yet | Does the action have a time, conditions, and a clear way to know it is done? | Set up a start reminder, reduce distractions, and add a way to ask for help |
| No next step after a setback | Can you describe what happened and give a reason to adjust or stop? | Track evidence and commitments. First look at where this attempt got stuck. Do not judge the person |
Do not create pressure through punishment, shame, forced payment, or sharing private matters publicly. Real responsibility includes delivering what was promised, explaining delays in advance, protecting data, and replying to participants. Stopping a direction that does not fit can be a good decision, but you need to explain why
Before you start
1. Read the anonymous case evidence page first. Separate what has happened, what is a hypothesis, and what still needs checking. WeChat articles are optional reading for after class. Class and self-study do not require sharing family experiences
2. Tell learners to open study.html first. Paper and pen work is fine. Reading markers are saved only in the person's browser. They do not mean the teacher sees progress, and they do not prove the task was done
3. Agree on one way to ask for help and a rough reply time. Learners only send non-sensitive summaries they are willing to share. No one needs to reply at all times. The page does not automatically submit or collect data
4. If you need to demo AI, prepare a synthetic scenario. If AI is not available, draw the flow, write the check, or try with paper cards. Do not wait for the tool
5. Do not install unknown tools in class, do not reverse-engineer apps, and do not enter real keys. Course materials can be read offline. WeChat and external sources need an internet connection to open
Five-part self-study path
| Section | Reading steps | Suggested time | What you keep |
|---|---|---|---|
| Start with care | 1 to 5 | 16 minutes | One scenario, one hypothesis about a need, mark the guesses |
| Think first, then use tools | 6 to 8 | 10 minutes | One trade-off, two resources, one AI task |
| Try it to know | 9 to 11 | 11 minutes | One minimal check, one data responsibility |
| Understand the first feedback | 12 to 15 | 13 minutes | One invitation, feedback categories, start and recovery plans |
| Make the next step happen | 16 to 17 | 10 minutes | 48-hour action card and seven-day review time |
Independent learners do this first, then open the self-check hints. You can invite a partner to review each other's work, but you can finish it alone. Mark paper exercises as “practice,” not real user feedback.
60-minute online meeting plan
| Time | Screen share page | How to move forward | What learners keep |
|---|---|---|---|
| 00—05 | 1 to 2 | Quiet topic selection, explain case boundaries | One self-chosen scenario |
| 05—11 | 3 to 4 | From communication concerns to a minimal experience, distinguish problem, need, and solution | One rewrite |
| 11—16 | 5 | Activity 1, write a hypothesis independently, then self-check or share voluntarily | Action Card A |
| 16—22 | 6 to 7 | Explain trade-offs and reusable accumulation | One principle, two resources |
| 22—29 | 8 to 9 | AI delegation, device acceptance, and small blockers | One acceptance judgment |
| 29—33 | 10 | Activity 2, write down one minimal check | Action Card B |
| 33—37 | 11 | Storage, authorization, costs, and beta risks | One responsibility and boundary |
| 37—44 | 12 to 13 | Invitation, installation, task, follow-up, distinguish feedback | One feedback classification |
| 44—47 | 14 | Activity 3, review analysis and replica tools | One authorization and verification judgment |
| 47—50 | 15 | Self-disciplined start and recovery, responsibility after failure | One sticking point |
| 50—57 | 16 | Activity 4: Write an action card independently, self-check, then revise | Action Card C |
| 57—60 | 17 | Leave three sentences and a review time | Action, time, evidence |
Activities 1, 2, 3, 4, and the final writing take 22 minutes in total. Do not spend all of it explaining the case. In the session, you are not required to turn on your camera or share personal experiences. You can participate by text or speech.
Four activities: How to give advice
Activity 1 | From problem to need
First write the phenomenon, then rewrite it using “who, when, obstacle, desired change.” Circle one part that lacks evidence and add: “If I see this evidence, I will change my mind.” Self-check when studying alone; in the session, you may voluntarily share one point.
Teaching example: “Older people lack communication” is too broad. You can change it to a hypothesis: A speaker who does not want to type for long periods wants to speak when they have something to say, without first organizing it into an article, and decides their own pauses and sharing. Willingness to use and willingness to share must still be asked separately.
If there is no topic, use the teaching scenario “It is easy to forget to return borrowed items.” First draw the current process, ask about the most recent experience, and do not use leading questions like “Do you really need this?”
If a student says they are not interested in the case, you can say: “Choose a scenario you are willing to spend 20 minutes understanding, and see if there is a real problem.” You do not need to like Shiguang Huaxia, nor do you need to be the user yourself.
Activity 2 | AI says it is fixed, but the device is still silent
First write the expected and actual results, then list what is known, what is unknown, and the minimum check. For example, play a synthesized text independently, record if it triggers and if it returns an error, then decide which part to check. Volume, playback trigger, network, or service errors are only possible causes.
The teacher can ask: “If you say it is fixed again, what extra evidence do you have?” If you can add reproduction steps, the difference between expected and actual results, and a re-test after one change, the exercise is complete. Platform syntax is not tested.
Activity 3 | Tools can replicate, but what do you decide?
You can choose one behavior to judge: Write your own needs after observing public interactions; reuse components according to the license; extract the other party’s credentials or customer data without authorization; or claim an AI copy is usable without testing it.
Clearly state the object’s authorization, the available scope, the source, and the results you must verify yourself. REA and replica-skill are for reading only; do not install or run them. A tool license does not equal authorization for the target content, and repository notes do not equal tested success.
Activity 4 | Set a time for the next step
2 minutes to write the object and hypothesis, 2 minutes to write the action and evidence, 2 minutes to self-check, 1 minute to revise. If you have a partner, you may voluntarily review each other’s work. Prioritize filling in the columns useful for this session; you do not need to complete all tables.
Teaching example: “After dinner tomorrow, spend 20 minutes drawing two paper cards for borrowing and returning items. Ask one voluntary peer to try it once. I will only note where I get stuck. If no one is available, do a rehearsal and mark it as such.”
If you still write “build a platform,” “do some research,” or “make everyone happy,” keep narrowing it down. You can commit to just one observation. If the object refuses, respect the refusal and arrange another verification.
Self-exploration, self-discipline, and first principles
The Strengths and Sticking Points Observation Card is an optional 5–8 minute reflection before class. It is not validated by psychological measurement, is not graded, and does not categorize types. See the Self-Exploration page for the differences and usage boundaries of DISC, formal MBTI, and 16Personalities. This course does not administer commercial questionnaires, collect psychological results, or assign roles or judge abilities based on types.
First principles are practiced on pages 4 and 8. Break down goals, facts, real constraints, and hypotheses to check, then use small tests to compare options. Self-discipline is practiced on pages 15 and 16, focusing on starting cues, reducing one distraction, and recovering after interruptions. Both methods support thinking and action, but neither guarantees success.
Success stories and verifiable interview clips are on the methods page as optional extensions, so they do not crowd out the main course. Unverified details about schedules, wealth, or intense conversations are not treated as classroom facts. Learners are not judged by staying up late or maintaining streaks.
Different learning setups, the same evidence standards
| Project | Lower secondary | University students learning online in distributed settings |
|---|---|---|
| Starting point | Accessible campus, life, or interest scenarios | Accessible people and scenarios for each learner |
| First step | Paper cards, observation, manual demos, or a small AI prototype | Non-technical prototypes are also fine; add budget and recruitment details |
| Interaction | Can be done independently, or voluntarily peer-tested | Independent reading and self-checks, with online help if needed |
| Responsibility | One small action, with adult support if necessary | Action, time, evidence, cost limit, and help plan |
| After seven days | One observation or trial, one revision | Aim for feedback from two independent people; if not possible, state this clearly |
| Long-term direction | Can be expanded later; no commitment to start a business is required | The year is a vision; decide the next step each week based on evidence |
Do not guess why a person acts or does not act based on ethnicity, region, or education. First look at the current task, resources, and existing experience, then discuss what kind of help is needed.
How to review a project
The main learning goal is application; the supporting goal is analysis. Each item is scored 0–2, for a total of 10 points. This evaluates only the current project and its evidence, not the person.
| Project | 0 points | 1 point | 2 points |
|---|---|---|---|
| Audience and context | Vague | Specific, but all guesses | Distinguishes observation from hypothesis |
| Values and trade-offs | Lists features only | Has principles, but they did not influence choices | Principles explain one trade-off |
| Validation design | Demo only | Has tasks, but criteria are unclear | Has a minimum task and judgment clues |
| Action plan | Wishes only | Has actions, but timing or evidence is unclear | Actions, timing, evidence, and backup plans are complete |
| Responsibility to others | Not considered | Verbal reminder | Explain consent, minimal recording, and access scope |
A score of 6 can serve as a reference for prioritising additional guidance, but it does not represent commercial validation. If the score is low, return to the missing parts; do not assign ability labels. Going forward, focus only on whether action started, what evidence was brought back, and how it was revised. A reading completion mark or a single emotional statement cannot prove long-term change.
A simple format for online feedback
Learners can manually send four sentences: “I want to understand…; I tried…; I noticed…; I’m stuck at…, and I’d like to ask…” Do not paste private chats, identity information, or real recordings. The page does not automatically send action cards, reflection cards, or reading progress to the teacher.
The teacher first responds to one specific question, then suggests one minimal action. In online meetings, prioritise discussing real sticking points and the next verification step; there is no need to read the webpage word for word. Response timing is agreed upon by both parties; immediate replies are not required.
Images, screen sharing, and printing
The site uses three background images and the AHALab Logo. The scenes are AI-generated teaching visuals, not real classroom settings. 19 diagrams show relationships and contain no fabricated statistics. Backgrounds and watermarks remain low-distraction; on mobile, tap images to enlarge.
For screen sharing in meetings, you can flip through pages one by one and expand key points as needed; N displays presenter tips. The webpage can also be read sequentially without a script. Printing retains diagrams and brand identifiers while removing large decorative backgrounds. The self-study version, action cards, and reflection cards can all be read offline; external source links require a network connection.
How to operate when presenting diagrams
When first entering a page, the diagram expands once to show relationships. You can pause when needed; press R to replay, or turn off animations and speak directly. Press S to open the presenter window, which shows suggested narration, reminders, the next page, and a timer for the current page. In online meetings, share only the main screen-sharing window. If the popup is not open, N still shows in-page tips.
Press F to request fullscreen; Home / End jumps to the first or last page. If the system setting reduces motion, view the static image directly; the final state is also shown before printing. The order of reading diagrams only helps understand relationships; it does not indicate evidence strength, user numbers, or project progress.