Voice AI technology has yet to achieve the breakthrough impact seen with ChatGPT, according to executives in the field. Despite billions of dollars invested in startups developing voice AI models and tools, industry leaders say the technology still faces challenges in natural conversation flow and fast reasoning. PolyAI’s CTO Shawn Wen shared these views at the HumanX conference last month, highlighting the gap between current capabilities and user expectations, according to techcrunch.com.
Shawn Wen noted that while full-duplex models, which can speak and listen simultaneously, represent a significant milestone, the next hurdle is improving reasoning speed to make conversations feel natural. Wen emphasized that AI agents in customer service must avoid sounding robotic to build caller confidence. He explained that once voice AI agents engage users for a few turns effectively, customers may increasingly trust them to resolve issues without human intervention, techcrunch.com reported.
The voice AI sector has attracted substantial investment across various applications, including enterprise customer service, meeting note-taking, and AI-powered dictation. However, experts like Otter’s CMO Alex Gay point out that features such as speaker identification and intent capture remain critical for progress. The current state of voice AI contrasts with the rapid adoption and impact of text-based AI models like ChatGPT, underscoring the complexity of voice as an interface, per techcrunch.com.
PolyAI’s CTO Shawn Wen’s remarks at the HumanX conference in late September provide a concrete timeline for the industry’s expectations, marking a focus on improving reasoning speed and conversational naturalness as key next steps for voice AI development.