Why you must fact-check AI responses in AI for Everyone - Visual Breakdown

Concept Flow - Why you must fact-check AI responses

User asks AI a question

↓

AI generates an answer

↓

Answer may contain errors or outdated info?

Yes→User fact-checks answer

↓

User finds correct info

↓

User trusts answer

↓

User uses info safely or corrects mistakes

This flow shows how AI answers can have mistakes, so users should always check facts before trusting or using the information.

Execution Sample

AI for Everyone

User: "Is the Earth flat?"
AI: "No, the Earth is round."
User: "Let me check sources."
User: "Confirmed, Earth is round."

A user asks AI a question, AI answers, user verifies the answer by checking trusted sources.

Analysis Table

Step	Action	AI Response	User Reaction	Result
1	User asks: Is the Earth flat?	No response yet	Waiting	Question posed
2	AI answers: The Earth is round.	The Earth is round.	Reads answer	Answer received
3	User checks trusted sources	N/A	Looks up facts	Verification in progress
4	User finds sources confirming Earth is round	N/A	Confirms AI answer	Trust established
5	User uses info safely	N/A	Applies knowledge	Correct use of info
6	If AI answer was wrong, user would correct info	N/A	Rejects wrong answer	Avoids misinformation

💡 User finishes fact-checking and decides to trust or correct AI answer

State Tracker

Variable	Start	After Step 2	After Step 4	Final
AI_Response	None	"The Earth is round."	"The Earth is round."	"The Earth is round."
User_Trust	No	Maybe	Yes	Yes
User_Action	Ask question	Read answer	Check sources	Use info or correct

Key Insights - 2 Insights

Why can't I trust AI answers blindly?

AI can make mistakes or have outdated info, so as shown in step 3 of the execution_table, users must check trusted sources to confirm.

What happens if AI gives wrong info?

As in step 6, the user should reject the wrong answer and find correct info to avoid spreading mistakes.

Visual Quiz - 3 Questions

Look at the execution_table at step 2, what does the AI respond?

A "I don't know."

B "The Earth is flat."

C "The Earth is round."

D "The Earth is a cube."

💡 Hint

Check the 'AI Response' column at step 2 in the execution_table.

At which step does the user confirm the AI's answer by checking sources?

A Step 4

B Step 1

C Step 3

D Step 6

💡 Hint

Look at the 'User Reaction' column where user finds sources confirming the answer.

If the AI gave wrong info, what should the user do according to the flow?

A Trust the AI anyway

B Fact-check and correct the info

C Ignore the answer and do nothing

D Ask AI again without checking

💡 Hint

Refer to step 6 in the execution_table and the key_moments about rejecting wrong answers.

Concept Snapshot

AI answers questions but can be wrong or outdated.
Always fact-check AI responses using trusted sources.
If AI info is correct, trust and use it safely.
If wrong, correct it to avoid mistakes.
Fact-checking protects you from misinformation.

Full Transcript

When you ask an AI a question, it gives an answer. But this answer might have mistakes or be outdated. So, you should always check trusted sources to confirm if the AI's answer is right. If you find the answer is correct, you can trust and use it safely. If it is wrong, you should correct it to avoid spreading wrong information. This process helps you use AI safely and avoid mistakes.

Practice

1. Why is it important to fact-check answers given by AI?

easy

A. Because AI always gives the most recent news updates.

B. Because AI can sometimes provide incorrect or outdated information.

C. Because AI never makes mistakes.

D. Because AI answers are always opinions, not facts.

Solution

1. Step 1: Understand AI's nature

   AI generates answers based on patterns in data but can include errors or outdated facts.

2. Step 2: Importance of verification

   Fact-checking ensures the information is accurate and trustworthy by comparing with reliable sources.

3. Final Answer:

   Because AI can sometimes provide incorrect or outdated information. -> Option B

4. Quick Check:

   AI answers may be wrong = A [OK]

Hint: Remember AI can err; always verify facts [OK]

Common Mistakes:

• Assuming AI is always correct
• Believing AI updates instantly
• Ignoring the need for trusted sources

2. Which of the following is the correct way to fact-check an AI response?

easy

A. Ignore the AI answer and guess the information yourself.

B. Trust the AI answer without checking.

C. Ask the AI the same question repeatedly until the answer changes.

D. Compare the AI answer with information from trusted websites or books.

Solution

1. Step 1: Identify proper fact-checking method

   Fact-checking means verifying information by consulting reliable and trusted sources.

2. Step 2: Evaluate options

   Only Compare the AI answer with information from trusted websites or books. describes comparing AI answers with trusted websites or books, which is correct.

3. Final Answer:

   Compare the AI answer with information from trusted websites or books. -> Option D

4. Quick Check:

   Fact-check by trusted sources = A [OK]

Hint: Always verify AI info with trusted sources [OK]

Common Mistakes:

• Blindly trusting AI answers
• Repeating questions hoping for different answers
• Ignoring reliable sources

3. An AI gives this answer: "The capital of Australia is Sydney." What should you do next?

medium

A. Check a trusted source to confirm the capital of Australia.

B. Ignore the question because AI answers are always wrong.

C. Assume the capital is Canberra because it sounds right.

D. Accept the answer as correct because AI is reliable.

Solution

1. Step 1: Recognize possible error in AI answer

   The AI says Sydney is the capital, but this is a common mistake; the actual capital is Canberra.

2. Step 2: Verify with trusted sources

   To be sure, check a reliable source like an official government website or encyclopedia.

3. Final Answer:

   Check a trusted source to confirm the capital of Australia. -> Option A

4. Quick Check:

   Verify AI facts with trusted sources = D [OK]

Hint: Double-check surprising AI facts with trusted sources [OK]

Common Mistakes:

• Believing AI without verification
• Guessing answers based on intuition
• Ignoring fact-checking altogether

4. You asked an AI for the latest COVID-19 statistics, but the data seems outdated. What is the best way to fix this?

medium

A. Ignore the outdated data and share it anyway.

B. Refresh the AI chat and ask again without checking other sources.

C. Check official health websites or news sources for the latest statistics.

D. Assume the AI will update automatically after some time.

Solution

1. Step 1: Identify the problem with AI data

   AI responses may not have real-time updates, so data can be outdated.

2. Step 2: Find reliable, current sources

   Official health websites or trusted news outlets provide the latest verified statistics.

3. Final Answer:

   Check official health websites or news sources for the latest statistics. -> Option C

4. Quick Check:

   Use trusted sources for current data = B [OK]

Hint: Use official sites for up-to-date info, not just AI [OK]

Common Mistakes:

• Sharing outdated info without checking
• Relying only on AI refresh
• Assuming AI updates instantly

5. You want to use AI to help write a research report. How can you ensure the information is accurate and trustworthy?

hard

A. Use AI answers as a starting point, then verify all facts with trusted sources before including them.

B. Copy all AI answers directly into your report without checking.

C. Only use AI answers if they sound believable to you.

D. Avoid using AI entirely because it is always wrong.

Solution

1. Step 1: Understand AI's role in research

   AI can provide helpful information but may include errors or outdated facts.

2. Step 2: Combine AI with fact-checking

   Use AI as a tool to gather ideas, then verify every fact with trusted sources before using it in your report.

3. Final Answer:

   Use AI answers as a starting point, then verify all facts with trusted sources before including them. -> Option A

4. Quick Check:

   Verify AI info before use = C [OK]

Hint: Always fact-check AI info before using in reports [OK]

Common Mistakes:

• Copying AI answers without verification
• Trusting AI based on intuition
• Avoiding AI completely without reason