Voice AI Execs Say the Technology Hasn't Had Its ChatGPT Moment Yet

Voice AI has attracted billions of dollars in investment and a steady stream of new models that claim to sound and converse like humans, but executives from PolyAI and Otter say the technology still has not had its ChatGPT moment. The comments came during on-stage conversations at the HumanX conference last month.

Full-Duplex Models Are Here, Reasoning Speed Is Not

Shawn Wen, CTO of enterprise voice AI platform PolyAI, said the industry has reached the milestone of developing full-duplex models that can speak while listening to a caller. The next challenge, he argued, is making reasoning fast enough that models can fetch answers quickly and keep conversations feeling natural. Wen added that customer service agents should not sound robotic and need to give callers confidence that problems can be solved. Once the voice is good enough for the first two or three turns of a conversation, he said, customers build confidence over time and stop feeling the need to reach a human.

Meeting Bots Need More Than Accurate Transcripts

Alex Gay, CMO of meeting notetaker Otter, said speaker identification, intent capture, and connecting transcriptions with organizational knowledge are key steps toward automation. Otter is also working on digital twins that could represent people in meetings, and Gay said the output voice must carry the same emotive expressions as talking to a human. The best meetings involve debate and strategic discussion underpinned by relationships, he noted, and without that an avatar is just a question and answer chatbot.

Transparency and Trust

Both executives flagged accuracy as a blocker. Wen said automatic speech recognition models often miss important keywords, which undermines the capture of full context. Gay agreed, saying transcription was never the end point for Otter but the layer on which productivity gains depend, and that flawed transcripts lead to flawed follow-up actions and lost trust in the platform. On transparency, Gay said tools should tell customers when they are being recorded or talking to AI, and Otter wants to notify everyone in a chat that a meeting is recorded even when its bot is not present. Wen said it is equally important in enterprise calls to establish that people are talking to an AI.

Sources: Techcrunch Ai