Google Assistant will change. At least its guts.
Google Assistant gained AI skills from its chatbot Bard late last year, enabling it to understand photos and use data from emails, papers, and more. Gemini, the company’s most recent GenAI software, may now power Google Assistant’s conversational abilities.
A new Gemini app for Android (which replaces the Google Assistant app) and an upgraded Google app for iOS (coming in weeks) will provide users with a Gemini-powered Google Assistant experience. Mobile-only Google Assistant products like Nest and Home speakers and displays won’t alter.
In a press event, Google Assistant VP Sissie Hsiao told Eltrys that the overlay fronts the Gemini models and offers contextual recommendations.
“This overlay allows you to do some really interesting things with Gemini,” she added. You may produce a caption from a photo beneath the overlay. Bringing up the overlay helps you comprehend and ask questions about the article you’re reading.
Images, text, and voice instructions may be sent to the Gemini-powered Assistant with a corner swipe, power button touch, or “Hey Google” hotword, depending on the device. For instance, you may submit a plant photo with the instructions, “I just bought this plant; give me tips and YouTube videos to help me take care of it,” and the Gemini-powered Assistant will try to help.
If you’ve tested a Gemini model on the web using Bard, you know it can do GenAI tasks. The Assistant integration doesn’t alter anything. In the onboarding flow, Google asks, “Help me craft a text response to my friend who is stressed at work,” “Give me some ideas to surprise my concert-loving friend on their birthday,” and “Help me incorporate more plant-based options in my diet.” Gemini in Assistant can make pictures like on the web, although it’s unclear which model is used; Bard previously used Imagen 2.
If you’re linked to the same Google account on both platforms, Gemini-powered Assistant chats on mobile will cross over to web experiences, and vice versa.
If the Gemini-powered Assistant isn’t working for you, a menu screen in the Android Assistant and iOS Google app lets you go back to the previous models.
At this point, Gemini models aren’t replacing Google Assistant’s previous ones. Quick phrases, which allow you to bypass “Hey Google” for certain activities, will use the outdated models. Timers, calls, and smart home gadget control will also be included. On the settings page, Google states that applications with assistant integration, such as Google Maps, may use the old models even when the Gemini models are chosen as the default models.
The most powerful Gemini model, Gemini Ultra, doesn’t power Assistant for free, which may annoy veteran Google Assistant customers. It and its improved thinking, coding, and instruction-following skills are locked behind Gemini Advanced, a new offering under the $20-per-month Google One AI Premium Plan. The free Gemini-powered Assistant uses the less-capable Gemini Pro, which arrived in Bard in December to mixed reviews.
In fact, Google is pitching the Google One AI Premium Plan, an extension of Google One with a two-month free trial, as the sole way to get its newest GenAI capabilities via Assistant. Gemini Ultra upgrades will include multimodal capabilities, interactive coding, and data analysis, but users must subscribe.
For added value, the AI Premium Plan includes 2TB of storage and Gemini-powered capabilities in Gmail, Google Docs, Slides, Sheets, and more. That may not be enough to sway Google Assistant customers who see the AI Premium Plan as a way to charge for free model upgrades.
Americans may now use the Gemini-powered assistant in English. Next week, new sites in Japanese and Korean will be added.
