These execs think voice AI hasn’t reached its ChatGPT moment yet

The current state of voice AI, while showing progress in areas like full-duplex communication, struggles with the underlying speed of reasoning and the accuracy of automatic speech recognition (ASR). PolyAI's Shawn Wen highlights that models need to fetch answers quickly for conversations to feel natural and build user confidence, particularly in customer service settings where a robotic sound can deter engagement.

A critical barrier to voice AI's effectiveness lies in its ability to accurately identify speakers and capture the full context of a conversation. Otter's Alex Gay notes that if the initial transcription, or ASR, is flawed, all subsequent actions and productivity gains derived from it become unreliable. This directly impacts user trust and the utility of applications like meeting note-takers.

The development of voice AI is not merely about sounding human but about delivering functional reliability. Both executives emphasize that the core issue is not just voice quality but the foundational accuracy of understanding and processing spoken language. This means continuous improvement in ASR models is essential, as downstream applications depend heavily on this foundational layer to avoid errors and maintain user confidence.

Transparency also emerges as a practical consideration for voice AI tools. Companies like Otter and PolyAI acknowledge the importance of clearly informing users when they are interacting with an AI or being recorded. This commitment to openness is presented as a way to build trust, especially as these technologies become more integrated into professional and customer service environments.