As part of the ongoing story, up to now, the voice modes for ChatGPT, Claude, and Gemini have had to take turns as they listen and talk, making for herky-jerky conversations with lengthy diatribes punctuated by abrupt interruptions. AI voice modes are so bad that I typically don’t bother.
According to the latest update, but we have seen some AI startups offering more natural-sounding voice modes, including some that actually listen while they talk, and now OpenAI is debuting an upgraded chat mode that’s also designed to be a much better listener.
The report highlights that rolling out now to ChatGPT users, GPT-Live (which I haven’t had the chance to try yet) boasts a key capability: a full-duplex architecture, which allows the model to both listen and talk at the same time.
The report highlights that that means when I talk, you listen — “Hey, good buddy, what’s your 20? Over” — and when you talk, I listen — “I’m at Dunkin Donuts on I-5, over.”. Previous ChatGPT voice modes were turn-based, similar to old-school CB radios.
Industry observers note that and because it “continuously processes input while generating output,” GPT-Live is better at deciding when to talk, interrupt, or — crucially — just shut up and listen. With GPT-Live (which is coming out in two versions, GPT-Live-1 and GPT-Live-1 mini), ChatGPT now employs a full-duplex configuration like on a standard phone call, meaning I can (if you don’t mind too much) interrupt or even talk over you.
According to the latest update, aside from being able to talk and listen at the same time, GPT-Live can also deploy agents to conduct web searches while it’s talking, thus eliminating the “searching the web” pauses on the old ChatGPT voice mode when it was looking up something.
In a fresh development, aI voice startup Thinking Machines has demoed voice “interaction” models that can also listen while they talk. There’s also Sesame AI, which has a voice mode capable of deploying search agents while it’s still talking. OpenAI’s GPT-Live models aren’t the first we’ve seen (and heard) with full-duplex capabilities.
In a fresh development, maybe ChatGPT will actually be worth talking to now. But for most of us, GPT-Live will be the first time we actually give full-duplex AI voice chats a try, and who knows?
According to the latest update, the fresh voice mode hasn’t landed on my ChatGPT account yet, but I’ll share my first impressions once it does. GPT-Live is rolling out to all ChatGPT users starting today, including free users.
As part of the ongoing story, his coverage of artificial intelligence interrogates the most recent LLMs, and how they can be used at work and at home to be best prepared for the AI revolution. “AI is going to change our lives sooner than we think,” Ben writes. “Our best way to adapt is by using it every day.” Ben has been a PCWorld author since 2014, and has covered everything from laptops to security cameras before launching PCWorld’s AI beat. Ben's articles have also appeared in PC Magazine, TIME, Wired, CNET, Men's Fitness, Mobile Magazine, and more. Ben holds a master's degree in English literature. Ben has been writing about consumer technology for more than 20 years, and now focuses his reporting on AI as it relates to the basic human experience.