The latest upgrade to ChatGPT, now rolling out to all users, is GPT-Live, a "new generation" of voice models designed to enhance interaction with the AI chatbot.
Unlike its predecessor, which was a relatively basic wrapper on top of the standard text input and output, GPT-Live promises to provide a more natural and fluid way to engage with the AI.
The new voice mode allows for the first time for the AI to think in the background while continuing the conversation, giving users extra space to pause when needed, and indicating it's still listening with phrases like "mhmm" or "yeah."
Users should find the upgrade on mobile and the web now, or very soon, with free users gaining access to GPT-Live-1 mini and those on paid plans able to access the even smarter GPT-Live-1 model.
OpenAI's goal is to make talking to ChatGPT feel like talking to a real person, and GPT-Live gets closer to achieving this.
Originally, interacting with the AI via voice required a specific model for speech-to-text, another for responding to the query, and another for text-to-speech.
GPT-Live combines all of this into a single AI model, allowing the chatbot to talk and listen at the same time, with users able to interrupt it as needed and responses being faster and more nuanced.
The new voice mode is also smarter at recognizing the difference between pausing mid-thought and finishing a query, recalculating several times a second whether to speak, continue listening, pause, interrupt, or invoke a tool.
One added benefit of the upgrade is that even complex work and deep thinking can be passed back to ChatGPT's servers in the background, while the conversation continues.
Users can also tell ChatGPT to take a beat or slow down, and visual responses have been improved, with pop-up cards for locations, weather forecasts, and sports scores.
Another new feature is the ability to ask GPT-Live to translate something into a foreign language as you speak, with a running translation in the other language without pauses or interruptions.
Additionally, improvements have been made in terms of ignoring background noise, such as background traffic or conversations happening nearby.
To access voice mode in the mobile app, users can tap the soundwave-style icon to the right of the prompt box, and the new mode looks a lot like the old one on the surface, but with the update, users should see "Live" at the top of the screen.
Right away, the upgraded voice mode feels more realistic and natural, with ChatGPT talking in a varied and expressive way, throwing in useful markers like "let me check" whenever it's looking something up.
There are three 'intelligence' levels to choose between, with varying levels of trade-off between the speed of the response and how detailed and accurate it is.
