OpenAI's new voice model wants you to talk over it (www.businessinsider.com)

OpenAI has introduced GPT-Live, a groundbreaking voice model that enhances ChatGPT's conversational capabilities by allowing it to listen and respond simultaneously, rather than waiting for users to finish speaking. During a livestream demonstration, OpenAI showcased how this full-duplex assistant can engage in real-time conversations, providing intermittent acknowledgments such as "mmhm" and "got it" while effectively processing multiple requests from users. This innovation aims to create a more human-like interaction with AI, making conversations feel seamless and dynamic. The significance of GPT-Live lies in its potential to revolutionize how users engage with AI, pushing the boundaries of voice assistant technology beyond traditional turn-based models. As noted by OpenAI's president, Greg Brockman, this update fosters a "much more natural way of interacting." The advancement not only improves user experience but also aligns with growing efforts in the AI community, such as similar initiatives by Thinking Machines, to develop interaction models that enable continuous audio, video, and text communication. This evolution suggests a shift toward more intuitive AI interfaces, addressing common criticisms of existing voice assistants and improving human-AI communication overall.