When it first came out, the Gemini app for Android was pretty bare bones. Since then, however, Google has added a bunch of neat features that make it a uniquely powerful assistant.
You can talk to the Google Assistant, in a way, if you count sending voice messages back and forth as talking. If you ask it for the weather or a movie review, it’ll give you the information. After every message, you have to press the mic button over and over. It’s good enough for setting an alarm, but the Google Assistant can't hold a conversation. Even if it could, you wouldn't find the conversation human-like. You might when you talk to Gemini.
If you open the Gemini app on your phone, you’ll find a small waveform button with a sparkle on it in the right corner of the screen. Tapping it launches the Gemini Live interface. Just press start to activate Gemini Live. Feel free to leave this screen now; Gemini Live will keep working in the background.
Originally, Gemini Live was only offered to paid subscribers. Now, it's in the regular free version of the app. It lets you talk to Gemini exactly how you would with a person. It’s built on a large language model like the regular Gemini, except it can speak in real-time with no lag, and you can interrupt it (or yourself) mid-conversation. You don’t have to press the mic button over and over. It’s just a free-flowing conversation.
I've been testing Gemini Live for a few weeks now, and it blew me away every time. If you’ve tried the ChatGPT Advanced Voice Mode, Gemini Live is a lot like that. There's nothing quite like this tech. I felt that “uncanny valley” quality that you have to experience yourself to really get it. Usually, when you’re texting Gemini, it takes a “just the facts” approach. When talking to Gemini Live, I felt as if it was actually listening to me. It understood me even when I trailed off or interrupted it.
You can set up Gemini Live by feeding the Google Assistant your voice and switching to Gemini as the default assistant. Gemini Live can talk to you in 10 different voices, including one with a British accent. You can change it later, too.
I tried discussing a movie with it, and it asked me a thoughtful or interesting question at every turn. It answered mine. And it almost sounded like Gemini had opinions about the movie. It even threw in some dialog filler like “huh,” “whoops,” and “hm” to create a more convincing impression. Gemini’s tone was also pretty informal. It didn’t sound robotic at all. This could be a really engaging way to learn more about a topic because you’re talking to a bot that knows everything and mimics human conversation.
I let it help me plan my day instead of just jotting down the tasks myself. We talked about the things I want to get done today. Gemini Live followed my lead and suggested the day's plan. I edited some of it just by talking to Gemini. At one point, it said I seemed overwhelmed, and I was. I followed its advice and took a minute to relax. We picked up the conversation where we left off. The whole time, it felt like talking to a real assistant.
Gemini Live is far from perfect, though. One hold-up is that Gemini Live can't access extensions yet (more on that in a few). So when I ask it to create a list of tasks, it sends the list in chat. Gemini Live also pauses pretty abruptly when you interrupt it mid-sentence. Gemini can hallucinate, too, just like other large language models, so that’s something worth keeping in mind. I think this incredible feature is still in the early stages of development, and it's going to eventually evolve into a full-fledged assistant that works with Google apps.
You can ask Google Assistant to connect with your apps to check your calendar or control Google Home. App integrations like those are rebranding to Gemini Extensions. You can wake up Gemini with “Hey Google,” too, and access the extensions with your voice. They work with text prompts, too.
For example, you could ask it to generate a to-do list or notes for you in Google Keep. Instead of starting from scratch, however, you could give it a prompt to generate the list automatically. You can also create new calendar events and manage existing ones using Gemini.
Ask it to play music using YouTube Music. You can start with a simple prompt like “play some music” or give it a specific album, artist, or track name—the same works with YouTube videos. More than just simple playback, you can ask complicated questions about the videos themselves. Google Maps is tightly integrated with Gemini. You can pull up map overlays without leaving the app and ask Gemini questions about navigation and places.
Google has connected the entire Google Workspace to work with Gemini Mobile. So, without opening Google Docs, Drive, Sheets, or Gmail, you can run a powerful search across the entire workspace, get AI-generated summaries, or create reminders and tasks out of your emails.
Of course, all these extensions can work in tandem, too. This is how Google describes it: you can ask Gemini to pull a recipe from your emails and add the ingredients to a Keep list with a single prompt. Gemini can also help you plan your next trip with the Hotels and Flights app integrations. It connects with Google Hotels in real time to fetch you hotels that match your specific prompt and itinerary. You can then find flights based on your itinerary using the Google Flights extension.
Beyond these extensions, Gemini can collaborate with Google Assistant to get everyday tasks done, too. You can ask it to set alarms, control smart home devices, or identify a song or tune. The Google Assistant also handles the Routines feature. Depending on when you’re reading this, Google might have added a lot more extensions. You can access the list by tapping your profile icon and selecting “Extensions” from the menu.
The Gemini app has a neat camera button that lets you take pictures and ask questions about them. You could ask it to caption an image, identify a plant, or tell you more about a medication. It’s surprisingly good at identifying and captioning both.
It can also talk to you about what's on your screen currently. With the Gemini overlay (you can pull up Gemini anywhere on your mobile phone by long-pressing the home button or holding the home gesture), you can quickly ask Gemini to identify what’s on the screen. It actually just takes a screenshot of your current screen and gives you more info about it. I found it pretty accurate the few times I had to use it. You can access this feature by opening the overlay and tapping the “Add This Screen” button. It uploads the screenshot to Gemini and lets you follow up using voice or text.
Gemini is not ready to fully replace the Google Assistant yet, but it is already much more powerful. With Google doubling down on generative AI for Android, I expect it to only grow more powerful in the coming years.
When you subscribe to the blog, we will send you an e-mail when there are new updates on the site so you wouldn't miss them.
Comments