Gemini Live keeps going unless you pause it, but some of Google's new Gemini AI features look really useful.

Google recently showed us its new Pixel lineup two months early at its Made by Google event, and the story was — surprise, surprise — all about AI. There’s the new, overly talkative Gemini Live voice assistant, a bunch of lame photo enhancement tools, and the obligatory copyright-skirting image-generation feature. Let’s take a look at the best and worst of Google’s latest AI toys.

“Google’s Gemini is an incredible piece of technology that’s built into Android 15. It has some really exciting features like Pixel Screenshots, which can search your screenshots for information. It’s a time-saving tool that’s both practical and forward-thinking. And Gemini Live voice chat, which could revolutionize real-time communication by adding AI to conversations,” Steven Athwal, founder and managing director of UK-based refurbished phone retailer The Big Phone Store, told Lifewire over email. “But Gemini’s AI image generation has me scratching my head. As someone who has always put ethics first in business, I can’t help but wonder about the copyright implications of this feature. It’s a fantastic tool, but the implications could be far-reaching, and not necessarily in a positive way.”

The promise of AI, of course, is an intelligent assistant that you can simply talk to, like the computers in every sci-fi utopia ever. As we’ve seen, the word “intelligence” in the name doesn’t exactly make it intelligent, but these LLMs (large language models) are better at figuring out what we’re saying than previous attempts. And according to The Verge’s Alex Cranz, Google’s Gemini Live is actually pretty good at figuring things out for you. When she tested the new assistant, it managed to give her an answer to a problem that had taken her and a friend five minutes to solve together in seconds.