Is Voice AI Really Improving or Are We Just Measuring It Wrong? 🤔

A voice model might perform well on traditional benchmarks, but does that mean it actually understands how humans communicate? Andrew Ettinger, CEO of Hume AI, explains how a decade of research led his team to develop a new approach to evaluating voice AI, using 1 million human judgments, 15,000 samples, and 60 models to measure performance against real-world outcomes. Watch the full episode! #VoiceAI #HumeAI #AIModels #AITesting #TheNeuron