Revolutionizing Conversational AI: GPT-Live Upgrade
The latest update to ChatGPT has been rolled out to all users, introducing a significant upgrade to the AI chatbot’s voice mode. Dubbed GPT-Live, this new generation of voice models promises to elevate the conversational experience to unprecedented levels.
Source: lifehacker.com
Previous iterations of voice mode in ChatGPT were relatively basic, serving as a wrapper for the standard text input and output. However, GPT-Live represents a substantial leap forward, allowing the AI to think in the background while continuing the conversation.
Source: lifehacker.com
One of the key features of GPT-Live is its ability to recognize when users pause mid-thought or finish their query. This is achieved through the model recalculating multiple times per second, determining whether to speak, continue listening, pause, interrupt, or invoke a tool.
Furthermore, the upgrade enables complex work and deep thinking to be passed back to ChatGPT’s servers in the background, while the conversation is ongoing. Users can also instruct the AI to take a beat or slow down, and visual responses have been improved, including pop-up cards for locations, weather forecasts, and sports scores.
Another exciting feature of GPT-Live is its ability to translate languages in real-time, allowing users to engage in conversations across language barriers without interruptions. Additionally, the model has been fine-tuned to ignore background noise, such as traffic or nearby conversations.
To access GPT-Live, users can tap the soundwave-style icon to the right of the prompt box in the mobile app. The new mode appears similar to the old one on the surface, but with the update, users will notice the ‘Live’ label at the top of the screen. For now, users can tap this to switch back to the older models.
During testing, the upgraded voice mode felt remarkably natural and realistic. ChatGPT would engage in conversations, throwing in useful markers like ‘let me check’ whenever it was looking something up, and hesitating and drawing out words at times.
While there were a few instances of speech glitching and conversation hanging, these issues were relatively minor and likely to be ironed out in future updates. Interruptions were handled well, with the AI pausing to acknowledge the user’s input before continuing its train of thought.
Users can also tweak the level of thinking ChatGPT puts into the new voice mode by adjusting the Intelligence sliders. There are three modes to choose from—Instant, Medium, and High—with varying levels of trade-off between response speed and accuracy.
