PolyAI CTO and Otter CMO say voice AI has not had its ChatGPT moment yet

Voice AI still has major hurdles before it reaches a breakout moment comparable to ChatGPT, according to executives from PolyAI and Otter speaking at the HumanX conference, as cited by TechCrunch. PolyAI CTO Shawn Wen said that even though full-duplex models have already been developed, the sector has not yet reached its defining turning point. In his view, the next technical hurdle is making inference fast enough for conversations to feel natural rather than delayed or mechanical. Otter CMO Alex Gay pointed to a different but related set of requirements for automation. He said speaker identification, intent capture, and the ability to combine those signals with organizational knowledge are key steps in building useful voice-based systems. He also stressed that generated speech needs to carry the same emotional expression people expect in a real human conversation. Without that, he said, the product remains little more than a question-and-answer bot. Both executives agreed that current voice AI systems still need improvement in understanding users and in transcription accuracy.