Voice AI Hasn't Hit Its ChatGPT Moment, Execs Say — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- PolyAI CTO Shawn Wen said voice AI has not yet had its 'ChatGPT moment' even with full-duplex models released, identifying faster reasoning as the next milestone needed for natural conversation flow.
- Otter CMO Alex Gay said speaker identification, intent capture, and integrating transcripts with organizational knowledge are the key steps toward automation, and the company is developing digital twins that could represent people in meetings.
- Investors have poured billions into voice AI startups spanning model makers, enterprise customer service providers, meeting note-takers, and AI-powered dictation tools, though Wen warned the technology still does not convincingly sound human.
- Otter's Gay said transcription accuracy is critical because inaccurate transcripts cause flawed downstream actions that erode user trust, while PolyAI's Wen added that ASR models frequently miss important keywords, undermining full context capture.
- Both executives stressed transparency: Otter notifies participants in meeting chats when recording is active even when its bot is absent, and PolyAI makes clear to callers that they are interacting with an AI rather than a human.
- The executives shared their views on stage at the HumanX conference last month.
Why it matters: The gap between voice AI hype and current capability means enterprise adoption remains gated by accuracy and trust — Otter risks losing platform credibility if ASR errors cascade into automated actions, and customer-service bots won't gain caller confidence until reasoning speeds up and AI identity is disclosed upfront.
Ask SkimNews
