Tavus Griffin Fooled 48%: Is Video Call Trust Over
Half the participants in a real world test could not tell the difference between a human and Tavus Griffin on video. The company describes Griffin as the first face to face human interaction model for conversational AI. The result raises a direct question about authenticity in live video. Trust is now a design problem.
💡 Frequently Asked Questions (FAQ)
- Q: What is Tavus Griffin?
- A: Tavus Griffin is marketed as the first face-to-face human interaction model for conversational AI, designed to simulate realistic human-like video conversations.
- Q: What does the 48% fooled rate mean?
- A: In a real-world test, 48% of participants could not distinguish between a real human and Griffin on video, showing AI faces are increasingly convincing.
- Q: Is video call trust over?
- A: Not over, but the results raise serious questions about authenticity in live video, making trust a critical design challenge for platforms and users.