
The AI Report Live with Liam Lawson Building Real AI Humans: The Future is NOW
5 snips
Nov 22, 2025 Quinn Favret, co-founder of Tavus and expert in human computing, and Hassaan Raza, also a co-founder focusing on product and culture, explore the fascinating intersection of AI and human interaction. They discuss pioneering 'human computing,' teaching AI to understand human nuances like tone and body language. The duo introduces Tavus Pals, proactive AI companions designed for real-world applications in therapy and education. They also emphasize ethical AI design and the importance of hiring a culturally diverse team to enhance accessibility and reduce biases.
AI Snips
Chapters
Transcript
Episode notes
Computers Should Meet Humans Where They Are
- Human computing flips the dynamic: machines learn human language and behavior so people don't have to learn machines.
- Quinn Favret says this lets anyone use computers like talking to a friend, removing technical barriers.
The Waltz: Multi‑Signal Conversational Models
- Teaching machines the 'waltz' of conversation requires perception, understanding, reaction, and action.
- Quinn Favret emphasizes modeling microexpressions, prosody, and contextual reactions for realistic AI humans.
Capture Rich Multimodal Signals
- Collect rich multimodal signals: facial expressions, audio prosody, and contextual reaction data for perception models.
- Use those inputs to build understanding, then map to realistic real-time reactions and actions.
