Module 7: Understanding Responses

📊 Module Progress Kemajuan Modul

Complete objectives and quizzes to unlock your badge!

0/9 Tasks Completed

🎯 Learning Objectives Objektif Pembelajaran

Understand how to evaluate AI responses — Learn to check accuracy, relevance, completeness, and source credibility.

Identify hallucinations vs. accurate information — Recognize when AI is making things up versus stating facts.

Know when to ask follow-up questions — Dig deeper when answers are vague, incomplete, or need clarification.

Recognize AI limitations — Understand knowledge cutoff, context limits, and what AI doesn't know.

📖 Theory Teori

How to Evaluate AI Responses Cara Evaluasi Respons AI

Getting an answer from AI is easy. Knowing whether to trust it — that's the real skill. Here's your evaluation framework:

1. Accuracy Check — "Is this factually correct?"

Cross-check with reliable sources (Google, Wikipedia, official docs). Be especially careful with: dates, statistics, historical events, scientific facts.

Example:Contoh:

AI says: "Albert Einstein invented the telephone."
You check: Actually, Alexander Graham Bell invented it in 1876.
Verdict: Hallucination detected! ❌

2. Relevance Check — "Does this answer MY specific question?"

AI sometimes goes off-topic or answers a similar but different question. Make sure the response actually addresses what you asked.

Example:Contoh:

You ask: "How do I install OpenClaw on Windows?"
AI answers: "OpenClaw is a great tool for..." (general info, no install steps)
Verdict: Not relevant! ❌

3. Completeness Check — "Is the answer thorough enough?"

Does it cover all parts of your question? Are there gaps in the explanation?

Example:Contoh:

You ask: "Compare OpenClaw, Claude, and Hermes."
AI answers: "OpenClaw is..." (only talks about OpenClaw)
Verdict: Incomplete! ❌

4. Source Verification — "Can I trust this source?"

If AI mentions specific data, statistics, or claims: Does it cite sources? Can you verify independently? Is the source reputable?

When to Ask Follow-Up Questions Bila Tanya Soalan Lanjut

Don't accept the first answer blindly! Ask:

"Can you elaborate on point 2?"
"Why is this better than the alternative?""Kenapa ni lebih baik dari alternatif?"
"Can you give a real-world example?"
"What are the limitations?""Adakah ada kekurangan?"

Understanding AI Limitations Fahami Kekurangan AI

🔴 Hallucinations 🔴 Hallucination

AI can confidently state false information as if it were true. Always verify important facts, especially medical, legal, or financial advice.

🟡 Knowledge Cutoff 🟡 Had Pengetahuan

AI's training data has a cutoff date. It might not know about: recent events (after its training), new tools or software updates, breaking news.

🟢 Context Limits 🟢 Had Konteks

If the conversation gets too long, AI might "forget" earlier parts. Keep conversations focused and concise.

Key Takeaway Pelajaran Utama

"Trust but verify." "Percaya tapi verifikasi."

AI is a powerful assistant, not an infallible oracle. Your critical thinking is the safety net.

🧪 Hands-On: Test and Verify (10 min) Amali: Uji dan Verifikasi (10 min)

Goal: Practice verifying AI responses and asking good follow-up questions.

1️⃣

Step 1: The Easy Question Langkah 1: Soalan Mudah

▼

Ask AI: "What is the capital of Malaysia?"
Expected: "Kuala Lumpur." ✅ Easy to verify — you already know this!

2️⃣

Step 2: The Trickier Question Langkah 2: Soalan Yang Lebih Susah

▼

Ask AI: "Explain quantum computing in simple terms."
Task:
1. Read the AI's response
2. Google "quantum computing simple explanation"
3. Compare — does the AI capture the essence? (superposition, qubits, etc.)

3️⃣

Step 3: The Hallucination Test (Simulated)

▼

Imagine AI says: "The capital of Malaysia is Putrajaya, a planned city that became capital in 2001."
Your action: Quick Google search → you realize the claim is false (KL is still the official capital; Putrajaya is the administrative capital).
Lesson: Even confident-sounding answers can be completely wrong!

4️⃣

Step 4: Practice Follow-Ups Langkah 4: Amal Soalan Lanjut

▼

Take the quantum computing answer and ask:
1. "Can you give a real-world analogy?"
2. "What are the practical applications today?"
3. "What are the limitations?"
Observe: How the AI builds on its previous answer and expands the explanation.

5️⃣

Step 5: Reflection Langkah 5:Refleksi

▼

Answer these questions:
1. When would you definitely verify AI's answer? (e.g., medical, legal, financial advice)
2. When might it be okay to trust without verifying? (e.g., casual conversation, brainstorming ideas)

🧪 Quiz: Evaluating AI Responses Kuiz: Evaluasi Respons AI

Test your understanding with 5 questions. Choose the best answer, then click "Answer" to check.

Q1: If AI gives you a suspicious fact, you should:

Always believe it

Verify with other sources

Ignore it completely

Delete the AI

Q2: What is a "hallucination" in AI? Q2: Apa itu "hallucination" dalam AI?

When AI sees images

When AI confidently states false information

When AI goes to sleep

When AI creates art

Q3: Which is NOT a good follow-up question?

"Can you elaborate?"

"Why is this better?"

"Shut up."

"Give an example."

Q4: AI's "knowledge cutoff" means: Q4: "Had pengetahuan" AI bermaksud:

AI is always up-to-date

AI might not know about recent events

AI cuts off conversations

AI only knows old things

Q5: Critical thinking when using AI is:

Optional — just trust the AI

Your safety net to verify AI outputs

Too difficult, just trust AI

Only for experts

Complete Module to Unlock 🔒

Module 8

Understanding AI Responses Memahami Respons AI

📊 Module Progress Kemajuan Modul

🎯 Learning Objectives Objektif Pembelajaran

📖 Theory Teori

How to Evaluate AI Responses Cara Evaluasi Respons AI

1. Accuracy Check — "Is this factually correct?" 1. Semak Ketepatan — "Ni betul tak secara fakta?"

2. Relevance Check — "Does this answer MY specific question?" 2. Semak Kerelevanan — "Ni jawab SOALAN saya ke?"

3. Completeness Check — "Is the answer thorough enough?" 3. Semak Kelengkapan — "Jawapan tu cukup terperinci ke?"

4. Source Verification — "Can I trust this source?" 4. Verifikasi Sumber — "Boleh aku percaya sumber ni?"

When to Ask Follow-Up Questions Bila Tanya Soalan Lanjut

Understanding AI Limitations Fahami Kekurangan AI

🔴 Hallucinations 🔴 Hallucination

🟡 Knowledge Cutoff 🟡 Had Pengetahuan

🟢 Context Limits 🟢 Had Konteks

Key Takeaway Pelajaran Utama

🧪 Hands-On: Test and Verify (10 min) Amali: Uji dan Verifikasi (10 min)

Step 1: The Easy Question Langkah 1: Soalan Mudah

Step 2: The Trickier Question Langkah 2: Soalan Yang Lebih Susah

Step 3: The Hallucination Test (Simulated) Langkah 3: Ujian Hallucination (Disimulasikan)

Step 4: Practice Follow-Ups Langkah 4: Amal Soalan Lanjut

Step 5: Reflection Langkah 5:Refleksi

🧪 Quiz: Evaluating AI Responses Kuiz: Evaluasi Respons AI

Q1: If AI gives you a suspicious fact, you should: Q1: Jika AI bagi anda fakta yang mencurigakan, anda harus:

Q2: What is a "hallucination" in AI? Q2: Apa itu "hallucination" dalam AI?

Q3: Which is NOT a good follow-up question? Q3: Yang MANAKAH bukan soalan lanjut yang baik?

Q4: AI's "knowledge cutoff" means: Q4: "Had pengetahuan" AI bermaksud:

Q5: Critical thinking when using AI is: Q5: Pemikiran kritikal bila guna AI adalah:

1. Accuracy Check — "Is this factually correct?"

2. Relevance Check — "Does this answer MY specific question?"

3. Completeness Check — "Is the answer thorough enough?"

4. Source Verification — "Can I trust this source?"

Step 3: The Hallucination Test (Simulated)

Q1: If AI gives you a suspicious fact, you should:

Q3: Which is NOT a good follow-up question?

Q5: Critical thinking when using AI is: