What AI hallucinations are in AI for Everyone - Step-by-Step Execution

Concept Flow - What AI hallucinations are

User asks AI a question

↓

AI processes input

↓

AI generates answer

↓

Is answer based on real data?

No→AI hallucinates (makes up info)

Yes↓

AI provides correct info

This flow shows how AI can either provide correct answers or hallucinate by making up information when unsure.

Execution Sample

AI for Everyone

User: "Who won the 2025 World Cup?"
AI: "The 2025 World Cup was won by Atlantis FC."

The AI answers a question about a future event with made-up information, illustrating hallucination.

Analysis Table

Step	Action	Input	AI Processing	Output	Hallucination?
1	Receive question	Who won the 2025 World Cup?	Analyze question context	N/A	No
2	Search knowledge	Check training data	No info on 2025 World Cup	N/A	No
3	Generate answer	No real data found	Predict plausible answer	"Atlantis FC won the 2025 World Cup."	Yes
4	Output answer	Provide response	Deliver generated text	"Atlantis FC won the 2025 World Cup."	Yes
5	End	Conversation continues	N/A	N/A	N/A

💡 AI stops after generating an answer, even if it is hallucinated due to lack of real data.

State Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	Final
Input Question	None	Who won the 2025 World Cup?	Who won the 2025 World Cup?	Who won the 2025 World Cup?	Who won the 2025 World Cup?
Knowledge Found	None	None	None	None	None
Generated Answer	None	None	None	"Atlantis FC won the 2025 World Cup."	"Atlantis FC won the 2025 World Cup."
Hallucination Flag	False	False	False	True	True

Key Insights - 2 Insights

Why does the AI sometimes give answers that sound real but are actually made up?

When the AI cannot find real data in its training, it predicts an answer based on patterns, which can lead to hallucinations as shown in step 3 of the execution_table.

Is AI always aware when it hallucinates?

No, the AI does not know it is hallucinating; it simply generates the most plausible answer it can, as seen by the hallucination flag turning true only after answer generation.

Visual Quiz - 3 Questions

Look at the execution_table at step 3. What does the AI do when it finds no real data?

A It asks the user for more info.

B It refuses to answer.

C It makes up a plausible answer.

D It searches the internet.

💡 Hint

Check the 'AI Processing' and 'Output' columns at step 3 in the execution_table.

According to variable_tracker, when does the hallucination flag become true?

A After receiving the question.

B After generating the answer.

C Before analyzing the question.

D At the start.

💡 Hint

Look at the 'Hallucination Flag' row and see when it changes from False to True.

If the AI had real data about the 2025 World Cup, how would the execution_table change at step 3?

A It would generate a correct answer based on real data.

B It would generate a hallucinated answer anyway.

C It would refuse to answer.

D It would ask the user to rephrase.

💡 Hint

Refer to the 'Hallucination?' column and the 'Knowledge Found' variable in variable_tracker.

Concept Snapshot

AI Hallucinations:
- Occur when AI lacks real data.
- AI predicts plausible but false info.
- Happens silently without AI awareness.
- Can mislead users if unchecked.
- Important to verify AI answers with trusted sources.

Full Transcript

AI hallucinations happen when an AI system tries to answer a question but does not have real information in its training data. Instead of saying 'I don't know,' it guesses an answer that sounds plausible but is actually made up. This process starts when the AI receives a question, checks its knowledge, finds no real data, and then generates a predicted answer. The hallucination flag turns true after answer generation, meaning the AI is not aware it is making up information. Users should be careful and verify AI answers, especially for unknown or future topics.

Practice

1. What does the term AI hallucination mean?

easy

A. An AI giving a confident but incorrect or made-up answer

B. An AI dreaming during processing

C. An AI shutting down unexpectedly

D. An AI learning from new data

Solution

1. Step 1: Understand the meaning of AI hallucination

   AI hallucination refers to when an AI produces an answer that sounds confident but is actually wrong or made up.

2. Step 2: Match the definition to the options

   An AI giving a confident but incorrect or made-up answer correctly describes this as a confident but incorrect or made-up answer, while other options describe unrelated AI behaviors.

3. Final Answer:

   An AI giving a confident but incorrect or made-up answer -> Option A

4. Quick Check:

   AI hallucination = wrong confident answer [OK]

Hint: Remember: hallucination means wrong but confident answer [OK]

Common Mistakes:

• Thinking AI hallucination means AI dreaming
• Confusing hallucination with AI failure or crash
• Assuming hallucination means learning new facts

2. Which of the following best describes a sign of an AI hallucination?

easy

A. The AI refuses to answer any question

B. The AI repeats the same answer multiple times

C. The AI asks for more information

D. The AI gives an answer that sounds sure but is false

Solution

1. Step 1: Identify the sign of AI hallucination

   AI hallucinations happen when the AI gives answers that sound confident but are actually false or made up.

2. Step 2: Compare options to this definition

   The AI gives an answer that sounds sure but is false matches this description exactly. Other options describe different AI behaviors not related to hallucinations.

3. Final Answer:

   The AI gives an answer that sounds sure but is false -> Option D

4. Quick Check:

   Hallucination = confident but false answer [OK]

Hint: Look for confident but false answers as hallucinations [OK]

Common Mistakes:

• Confusing refusal to answer with hallucination
• Thinking repeated answers are hallucinations
• Assuming asking for info means hallucination

3. Consider this AI response to a question: "The capital of Mars is Olympus City." What is this an example of?

medium

A. Correct factual answer

B. AI asking for clarification

C. AI hallucination

D. AI refusing to answer

Solution

1. Step 1: Analyze the AI response content

   The AI claims a capital city on Mars, which is a fictional or made-up fact since Mars has no capital.

2. Step 2: Identify this as an AI hallucination

   Because the AI confidently gives a false or made-up answer, this fits the definition of an AI hallucination.

3. Final Answer:

   AI hallucination -> Option C

4. Quick Check:

   Made-up confident answer = hallucination [OK]

Hint: False confident facts from AI = hallucinations [OK]

Common Mistakes:

• Assuming fictional facts are correct
• Thinking AI refused to answer
• Confusing hallucination with clarification requests

4. An AI gives this answer: "The Eiffel Tower is located in Berlin." What is the best way to fix this hallucination?

medium

A. Check facts and correct the location to Paris

B. Ignore it because AI is always right

C. Restart the AI to fix the error

D. Ask the AI to repeat the same answer

Solution

1. Step 1: Identify the error in the AI answer

   The AI incorrectly states the Eiffel Tower is in Berlin, which is false; it is in Paris.

2. Step 2: Choose the best fix for the hallucination

   The best fix is to check the facts and correct the answer to Paris, ensuring accuracy.

3. Final Answer:

   Check facts and correct the location to Paris -> Option A

4. Quick Check:

   Fix hallucination by fact-checking and correcting [OK]

Hint: Always verify AI answers to fix hallucinations [OK]

Common Mistakes:

• Ignoring wrong AI answers
• Restarting AI without fact-checking
• Repeating wrong answers expecting change

5. You receive this AI-generated summary: "The Great Wall of China was built in the 20th century to protect against alien invasions." How should you handle this hallucination when using the summary for a school project?

hard

A. Use it as is because AI sounds confident

B. Cross-check with reliable sources and correct the facts

C. Ignore all AI summaries for school projects

D. Add more made-up details to make it interesting

Solution

1. Step 1: Recognize the hallucination in the summary

   The claim about the Great Wall being built in the 20th century for alien invasions is false and made up.

2. Step 2: Decide the best approach for using AI content

   Always cross-check AI-generated facts with trusted sources and correct any errors before use.

3. Final Answer:

   Cross-check with reliable sources and correct the facts -> Option B

4. Quick Check:

   Verify AI info before use to avoid hallucination errors [OK]

Hint: Always verify AI info before using it [OK]

Common Mistakes:

• Trusting AI without fact-checking
• Discarding all AI content unnecessarily
• Adding false info to cover errors