Back to Home
Inside an LLM
Hallucinations
AI hallucination means the AI gives an answer that sounds confident but may be wrong.
You just learned how prompts guide AI responses. Now see what happens when AI is asked something it has no evidence for, and why a confident answer is not always a correct one.
This is a simple learning demo using predefined examples, not a real AI model.
Interactive Playground
85%
Low confidence
High confidence
Hallucination Risk
⬇️ Low
⚡ Medium
🔥 High
Live Visualization
💬
User Question
Waiting for question…
↓
📚
Evidence
No evidence checked yet
↓
🤖
AI Answer
Generate to see an answer…
Statistics
:
Evidence
85%
Confidence
:
Risk Level
:
Answer Type
:
Recommendation
How It Works
💬
Question
→
📚
Check
Evidence
Evidence
→
🤖
Generate
Answer
Answer
→
⚖️
Verify
Risk
Risk
🎓
Beginner Tips
2 tips
Confidence ≠ correctness. AI sounds certain even when it's wrong. Always verify important facts from a trusted source.
Grounding (RAG) reduces hallucinations, feeding the AI real documents constrains it to answer from evidence, not imagination.
💡
Key Takeaway
AI answers are safer when based on clear evidence. A confident answer is not always a correct answer.