Google is today beginning to roll out Gemini Live, a promised conversational version of the company’s generative AI. The tech is similar toChatGPT’s voice chat mode, letting people go back and forth withGeminiinstead of making one-off requests. Its functions aren’t fundamentally different – but users can interrupt the AI’s answers, and quickly follow up on requests without having to remind Gemini about the context.
Pixel 9: What’s changed, what hasn’t?
The Pixel 9 features a slightly larger display, an upgraded ultrawide camera, and several new AI tools.
Users can also switch between 10 different voices – male and female – and pause a conversation if you want to resume it later. Perhaps more importantly, Gemini Live can continue running in the background of Android, even when your phone or tablet is locked. When it’s foregrounded, it gets a fullscreen interface with special effects.
There are some major limitations at the moment. Gemini Live is only available in English, and you’ll also need an Android device signed in toGemini Advanced, which comes bundled with a Google One AI Premium subscription. That’s $20 per month, although you do of course get perks like 2TB of Google Drive storage. Support for iOS and other languages is promised “in the coming weeks.”
What can you do with Gemini Live?
Google gives examples like rehearsing for a job interview, learning how to de-stress, or asking how to not just make invisible ink but make it more interesting. Notably, though, the launch version of Gemini Live lacks multimodal input, which would let it do things like scan a sign or tell you about an on-camera object. That’s something Google is introducing to other aspects of its AI.
Multimodal support is coming to Gemini Live – but only sometime “later this year,” according to Google.
