Tutorial: Understanding AI Verification

Reading time: 2 minutes

Ever wondered if an AI is telling you the truth? GlyphChat solves this with real-time verification.

The Problem with AI

Traditional AI chatbots can:

• Make up facts (“hallucinate”)
• Sound confident while being wrong
• Mix real and fake information

You have no way to know what’s true.

The GlyphChat Solution

We check every AI response against Glyphmine—a database of 28,100+ verified concepts and 184,000+ relationships.

See It In Action

Example 1: Verified Fact

You ask: “What is water made of?”

AI says: “Water is made of hydrogen and oxygen, with the chemical formula H₂O.”

GlyphChat verifies:

• ✓ Water exists in our database
• ✓ Hydrogen exists in our database
• ✓ Oxygen exists in our database
• ✓ We found connections between them

Result: Verified (83% confidence)

Example 2: Unverifiable Claim

You ask: “What will the stock market do tomorrow?”

AI says: “Based on current trends, the market may rise 2%.”

GlyphChat verifies:

• ✗ Future predictions cannot be verified
• ✗ No factual basis in knowledge graph

Result: Warning (0% confidence)

Key Takeaways

1. Green badge = AI response aligns with verified knowledge
2. Warning = Take with a grain of salt, verify independently
3. Click badges to see exactly what was checked
4. Click concepts to explore the underlying knowledge

Try It Yourself

1. Go to GlyphChat
2. Ask a factual question like “What is the largest planet?”
3. Watch the verification badge appear
4. Click the badge to see details

Next Steps

• Learn to read confidence scores
• Understand verification badges