ServiceNow ML engineer Midam Kim argues that voice AI systems fail because they treat conversation as a pipeline rather than a dynamic linguistic joint activity across sounds, vocabulary, turn-taking, and shared mental models.
FireBrief
FireBrief