ChatGPT voice: has it finally learned to shut up?

Previously, ChatGPT talked like a waterfall, without waiting for its turn. Now, with the new GPT-Live-1 model, it's been trained to be more polite. Finally, we'll be able to finish a sentence without the AI cutting us off, a nice step forward for everyone's patience.

The main novelty is that the system now knows how to wait. If you pause, maybe to think or catch your breath, it will wait. It's no longer that annoying colleague who finishes your sentences, making conversations much less frustrating. Kundan Kumar, OpenAI's research lead, described this version as "more like talking to another person," quite a compliment for an algorithm.

It seems trivial, but an AI that knows how to listen changes everything. Until now, voice interactions were a mechanical back-and-forth, often interrupted. Now the experience becomes more like a real chat, with a more natural flow. Isn't that what we're looking for from a virtual assistant, a bit of humanity?

Smooth interaction means less stress for the user. You won't have to worry about formulating perfect sentences or speaking without pauses. The system adapts to your pace, making voice use much more pleasant. OpenAI introduced the GPT-Live-1 model specifically to enhance the naturalness of voice conversations.