💾 Auto-Save: Your work is automatically saved as you type. You can close the browser and come back later.
Week 3 · Student Packet
Poster Coaching Guide & Peer Feedback Sheet
Use Part 1 to sharpen your poster content. Use Part 2 to draft your answers. Use Part 3 to give structured peer feedback.
Team Name: _______________________
Date: _______________________
Part 1 — Coaching Guide
What's Happening This Week
Your team reads your draft answers out loud. The class gives feedback. You take notes, improve your answers, then submit the Google Form officially before the deadline. Your poster gets printed from that submission — so what you submit is what goes on the wall at the exhibition.
The Poster Workflow
Week 3 (today): Prepare draft answers → read them out loud → get feedback → note what to fix.
After Week 3: Finalize your answers → submit the Google Form officially (or edit your existing submission).
Before Week 4: Form deadline passes → teacher runs the poster generator → A3 posters printed for the exhibition.
The Judging Rubric
This is exactly how your project will be scored at the exhibition. Your poster answers map directly to these criteria.
Criteria
3 — Strong
2 — Developing
1 — Weak
Problem Clarity
Real, specific, compelling
Somewhat clear
Vague or generic
AI Solution + ML Type
Named, justified, clearly explained
Named but vague
Missing or incorrect
Data & Bias
Sources + bias type + specific mitigation
Partial — mitigation missing
Not addressed
Ethics & Impact
Specific harm named + safeguards explained
Acknowledged but not addressed
Not addressed
Limitations & Failures
2+ specific failure scenarios
Vague acknowledgment
"It will always work"
Future Work
Ambitious, specific, grounded
Generic ("add features")
Not included
Real-World Examples
Named + compared to your system
Named but not compared
Not included
Final score = 60% judge rubric + 40% audience ranked-choice vote at the exhibition.
Section-by-Section Coaching Tips
The Problem — Make them feel it
Not: "Healthcare is important in Lebanon." Instead: "Elderly patients in Lebanese households miss medication doses regularly because there is no reminder system their families can monitor remotely." Specific person, specific situation, specific gap.
The AI Solution — Name the ML type like it's obvious
Not: "We use AI to solve this." Instead: "We use a classification model that takes symptom inputs and outputs a recommended specialist category." ML type named. Input named. Output named. Why AI and not a simpler tool — addressed.
How It Works — Step by step, not vague
Walk through it like you're explaining to someone who has never used your system. Step 1: user does X. Step 2: system processes Y. Step 3: output is Z. Don't skip steps. Don't say "the AI analyzes the data" — what data? What analysis?
Data & Bias — Specific = credible
Name the bias type. "We might have sampling bias because our training data comes from one hospital in Beirut, which means rural patients are underrepresented." Then name the mitigation: "We would collect data from at least 3 different regions." Vagueness loses points. Specificity wins judges.
Ethics & Impact — Own the risks
"Our system is safe" is the worst answer. Give a real harm scenario: "If the model misclassifies a serious symptom as low-risk, the patient delays seeking help and their condition worsens." Then give a real safeguard: "We include a disclaimer that the system does not replace a doctor and flags any output as a suggestion only."
Limitations & Risks — Two concrete failure cases
Format: user does X → system outputs Y → consequence is Z. Example: "A user types symptoms in Lebanese dialect → the NLP model trained on MSA fails to parse correctly → outputs a wrong classification." Two of these. Specific. Real.
Future Work — Dream big, stay grounded
Not: "We would add more features." Instead: "In Phase 2, we would integrate with pharmacy databases to auto-flag drug interactions. In Phase 3, we would build a family monitoring dashboard." Specific phases. Specific additions. Shows you've thought beyond the MVP.
Real-World Examples — Compare, don't just list
Not: "Google Maps does something similar." Instead: "Google Maps uses traffic prediction similar to ours, but it is not optimized for Beirut's road network or local shortcuts. Our system trains specifically on Lebanese commute data." Name it, then say what makes yours different or more relevant.
Writing for a Poster — Important Constraints
Your answers go directly onto the A3 poster
Write in full sentences — not bullet points. The poster layout expects prose.
Keep each answer to 3–5 sentences maximum — it needs to fit in its section on the A3 sheet.
Avoid filler phrases like "In conclusion" or "As mentioned above" — every sentence should add information.
Read each answer aloud. If it sounds robotic or vague, rewrite it.
What Good Feedback Looks Like
✅ DO
Be specific: "Your bias section names the type but doesn't say how you'd fix it."
❌ DON'T
Be vague: "I didn't really understand the bias part."
✅ DO
Frame challenges as questions: "How would your system handle inputs in Lebanese dialect?"
❌ DON'T
Attack the idea: "This idea doesn't make sense."
✅ DO
Give a concrete suggestion: "Add a specific safeguard to your ethics section."
❌ DON'T
Give empty praise: "It was really good, I liked it."
✅ DO
Point to the rubric: "Your limitations section would score a 1 right now — you need two specific failure cases."
❌ DON'T
Be discouraging: "You're going to lose." Challenge the content, not the team.
Part 2 — Draft Your Answers
Write your draft answers here before class. This is what your team will read out loud for peer feedback. After feedback, finalize and submit the Google Form.
The Problem
What issue are you solving? Who has it? How bad is it? (3–5 sentences)
The AI Solution
What does your system do? What ML type does it use and why? (3–5 sentences)
How It Works
Step by step — how does the AI process input and produce output? (3–5 sentences)
Data & Bias
What data is needed? What specific bias type could appear? What's your mitigation? (3–5 sentences)
Ethics & Impact
Who could be harmed if it's wrong? What specific safeguards exist? (3–5 sentences)
Limitations & Risks
Give 2 specific failure scenarios: user does X → system outputs Y → consequence is Z. (3–5 sentences)
Future Work
What would you improve or expand given more time and resources? Be specific. (3–5 sentences)
Real-World Examples
Name at least one existing system similar to yours. How does yours differ or improve on it? (3–5 sentences)
Before You Read Out Loud — Self Check
Is the ML type named explicitly in the AI Solution answer?
Is a specific bias type named in the Data & Bias answer?
Are there two concrete failure scenarios in Limitations & Risks?
Does every answer stay within 3–5 sentences?
Does any answer still say "our AI is safe" or "we'll use good data"? Rewrite it.
Part 3 — Peer Feedback Sheet
Complete one per team you give feedback to. Use specific, constructive language.
Rubric Scores
Score each section as you listen. 3 = Strong, 2 = Developing, 1 = Weak.
Problem Clarity
AI Solution + ML Type
Data & Bias
Ethics & Impact
Limitations & Risks
Future Work
Real-World Examples
⭐ What worked — be specific
🔍 One challenge — frame as a question
💡 One concrete suggestion
Part 4 — Post-Feedback Action Items
After receiving feedback, list the top 3 things you will fix before submitting the Google Form.