Executives Say Voice AI Has Yet to Reach Its ChatGPT Moment

Despite a wave of excitement that voice is becoming the next major interface, investors have poured billions of dollars into voice AI startups spanning model development, enterprise customer service, meeting note‑taking, and AI dictation. Yet leading figures in the field argue the technology has not yet achieved the breakthrough moment that ChatGPT delivered for text.

Shawn Wen, chief technology officer at enterprise voice AI platform PolyAI, told the audience at last month’s HumanX conference that full‑duplex models—capable of speaking while listening—have become a milestone, but the real challenge lies ahead. He explained that the next step is to accelerate reasoning so answers can be fetched rapidly, making conversations feel natural. Wen added that customer‑service AI agents must avoid sounding robotic and should give callers enough confidence that their problems can be solved. “When the voice is good enough and the customer stays engaged for the first few exchanges, confidence builds, and over time they may feel they no longer need to speak to a human,” he said.

Alex Gay, chief marketing officer for meeting note‑taker Otter, emphasized that speaker identification, intent capture, and linking those insights with organizational knowledge are essential for automation. The company is also developing digital twins that could represent people in meetings. For such avatars to be effective, Gay said the synthesized voice must convey the same emotive nuances as a human participant. “The best meetings involve debate and strategic discussion underpinned by relationship. If an avatar cannot replicate that, it becomes just a Q&A chatbot,” he noted.

Understanding and Transparency in Voice AI

Wen and Gay agree that current voice AI often fails to grasp user intent, leading to inaccurate transcripts or flawed summaries. Wen pointed out that automatic speech recognition (ASR) models frequently miss critical keywords, undermining context capture. Gay echoed this concern, noting that Otter continuously refines its transcription engine because downstream actions depend on its accuracy. “If the original transcript isn’t reliable, every follow‑up action becomes flawed, eroding trust in the platform,” Gay said.

Beyond accuracy, the new generation of voice tools raises transparency questions. Companies should clearly inform users when they are speaking to an AI or being recorded. Otter plans to embed notifications in meeting chats even when its bot isn’t present, aiming to build trust among participants. PolyAI’s Wen also stressed the need to establish that callers are interacting with AI during enterprise conversations, ensuring users know the nature of the interaction.