Video Today, Google said it will add more Gemini AI features to Android smartphones - though what's said to be best of that functionality will be exclusive to its new Pixel 9 line of handhelds.

The Chocolate Factory unveiled this latest stuff at its Made By Google event, which unsurprisingly focused largely on the Android ecosystem and machine learning.

The web titan said its generative Gemini assistant for Android has been "completely rebuilt," and now incorporates the Gemini Flash 1.5 model, all in the hope of making the thing better. You know the drill with this kind of software: You interact with it with natural language, ask it things, it answers.

Only a few of these new features intended for all Android phones - ones that run Android 10 or newer - are available today. The Gemini overlay feature can now take a screenshot of an app and then answer questions about it; users can also ask it to generate images and then drop them immediately into Gmail and Google Messages. These are the prime examples of the "deep integrations" Google boasts of with Gemini and Android.

While this humble vulture was able to get both of these features to work just fine, personally, I have to wonder why I need Gemini to describe what I'm looking at. There's always the accessibility side, for sure, which Google's also boosted with additional functionality, described here.

Perhaps there's just a lot of people out there baffled by the apps they use or the pages they read and need a hand; maybe they'll find this feature helps then extract info from their screens (there's also text search support in the screenshot-taking app). This feature seems to cover YouTube, where again if you're not sure what you're looking at, the assistant might be able to help you pull out certain information from it.

"If you’re using YouTube, ask questions about what you’re watching," Google veep Sissie Hsiao suggested. "Let’s say you’re preparing for a trip abroad and have just watched a travel vlog - tap 'Ask about this video' and ask for a list of all the restaurants mentioned in the video - and for Gemini to add them to Google Maps."

A far more interesting feature is Gemini Live, which is supposed to be a voice-based assistant that you can have a real conversation with, even to the point where interrupting it is okay. The AI pal is supposed to come out today for Gemini Advanced subscribers (a $20 monthly plan) tho at least on my Samsung Galaxy S23, it's not yet available.

Hopefully it'll be real this time.

Further additions to Gemini will apparently allow users to ask the AI to dig through your emails and find what you're looking for, perhaps some ingredients for a recipe someone sent you, the ad biz proposes. Users will also be able to ask Gemini to generate playlists (with real songs, to be clear) and handle calendar stuff.

Notes from phone calls can be automatically generated by on-device artificial intelligence. "If you need information like an appointment time, an important address or a phone number to call back, turn on Call Notes and all the details and transcript will be available in the call log," says Google. "To protect privacy, Call Notes runs fully on-device and everyone on the call will be notified if you have activated the feature."

An "add me" function allows you to take two photos and merge them so that everyone in the pictures can be seen together, even therefore the ones taking the shot if someone swaps over, the goal here being that no one gets missed out. That's among other AI-based functionality seemingly available for Pixel 9 devices.

If you want to find out more, see the pages we've linked to or scrub through the full event below.

The Chocolate Factory promises it'll treat users' private data with respect, saying that "only Gemini can do all of this with a secure, all-in-one approach that doesn't require hand-off to a third-party AI provider you may not know or trust."

However, Google hasn't exactly got off to a good start with Gemini and privacy concerns.

You want the real goods? Get a Pixel 9

There's not that much on offer for the entire Android ecosystem so far, and for those that want much more, they'll have to get one of the four Pixel 9 models that launch from August 22.

Google promises Pixel 9 owners will get some fairly robust features: A text-to-image generator, a search engine for saved images, a weather app (but with AI!), comprehensive and easy photo editing, and higher quality calls. All of this runs locally on the Pixel 9's new Tensor G4 chip and through the Gemini Nano multimodal model.

That these are Pixel-exclusive features might be partially down to the fact that not all phones have the horsepower to run AI workloads on-device; the G4 is claimed to be more powerful than earlier iterations. But part of it is undoubtedly Google wanting its own killer AI features, and it also gets to say the Pixel 9 is more private than other Android phones thanks to its on-device AI capabilities.

The Pixel 9 lineup and its unique AI features are seemingly Google's answer to Apple Intelligence, which launched in beta with many similar features just a couple of weeks ago. It's up in the air which company's AI is better, but Cupertino does have a monopoly on AI-generated emojis.

The Pixel 9 doesn't come cheap, with the base model (the regular Pixel 9) starting at $799 with 128GB and 256GB storage capacities; 6.3in OLED screen; 4,700 mAh battery; software updates for seven years; and a Tensor G4 processor with 12GB of RAM.

Google also offers three Pro models: the $999 Pixel 9 Pro (as regular but with 128GB, 256GB, 512GB, or 1TB storage; 16GB RAM; LTPO OLED screen), the $1,099 Pixel 9 Pro XL (as Pro but with 5,060 mAh battery; 6.8in LTPO OLED screen), and the $1,799 Pixel 9 Pro Fold (second-gen foldable 8in screen device coming in September). As an added bonus, Pixel 9 owners get Gemini Advanced for a year, which would normally cost $240.

Interestingly, Google's adding what's dubbed satellite SOS support to Pixel 9 devices this year in the US on Android 15, when that arrives, and possibly other countries later. For America, users will get two years of service for free. The idea, like what Apple added to its phones, is that if you're in an emergency and can't get any regular cellular or Wi-Fi signal, your handheld can attempt to use satellite connectivity to raise the alarm to responders and select contacts for you. Handy if you're lost in the middle of nowhere.

Google additionally announced Pixel Watch 3, which starts at $349.99. It too of course has AI features, such as automatically setting a bedtime mode based on sleeping habits, providing recommendations on running better, and detecting when a user's heart has suddenly stopped, which can trigger an automatic call to 911.

Hopefully that last feature works correctly, because it would be really awkward to have to tell paramedics that you're not dead. ®

PS: The US Dept of Justice is reportedly mulling seeking the break-up of Google to combat its monopolies. That could involve hacking off its AdWords, Chrome, and Android divisions, or at least banning the biz from signing any more contracts of the kind that make it the default search or make it the exclusive app store provider on devices.

https://www.theregister.com//2024/08/13/google_gemini_ai_pixel/