Xperience podcast

When AI Fails

9.11.2026
0:00
42:06
Spol 15 sekunder tilbage
Spol 15 sekunder frem

Ninety-five percent reliability might win a demo, but at a million conversations it means fifty thousand failures. In this episode, host Melissa Dixon talks with Dan Klein, UC Berkeley professor and cofounder and CTO of Scaled Cognition, and Peter Graf, Chief Strategy Officer at Genesys, about why reliability is the foundation for AI that acts on a customer’s behalf.

You’ll hear why large language models can look flawless in a demo yet struggle to ship, why their errors are so hard to spot, and how large action models take a different approach by making guarantees about what a system will, and will never, do. You’ll also hear why leaders who hold back on AI leave real value on the table, and what it takes to build the confidence to move forward with agentic AI.

Ready to put these ideas into practice? Download the CX Buyer’s Guide: https://www.genesys.com/resources/cx-buyers-guide


Flere episoder fra "Xperience"