When Google introduced the Google Pixel 8 Pro earlier this year, the company made it clear that it was more of an AI phone than just a regular Android flagship. Now, Google is leaning more into that AI future with a pair of new features based on the in-house Gemini Nano model.
In order to bring Gemini Nano to the Pixel 8 Pro, Google built a new system called Android AI Core in Android 14. It’s the most compact and efficient AI model from Google, operating on Tensor mobile silicon natively. On-device operation means all processes happen locally, which ensures there’s little risk of sensitive information leaving your phone.
Most importantly, you need not be connected to the internet – or Google’s cloud – for these AI features to work. Talking about features, Gemini Nano offers tricks such as advanced text summarization, context-aware smart responses, and sophisticated proofreading and grammar-correction tools.
The first Gemini Nano-based feature on the Pixel 8 Pro is a new summarizer system in the Recorder app for Pixel phones. The app, which already excels at transcription, will now present a summary of the audio recording in the form of quick-to-absorb bullet points. GPT-based tools and apps such as Shortwave also offer their own take on summarization, but they can’t do so without an internet connection.
For now, however, the AI summarizer system in Recorder is limited to the English language. Unfortunately, we don’t know whether support for more languages – and older devices – will be added down the road. Interestingly, the app is now able to transcribe audio in 28 languages.
Next, we have a new smart reply system in the Gboard app. The feature is already live globally in WhatsApp, but more apps will add support for this predictive text response feature in Google’s keyboard app soon. Conversational awareness is the trick here, and since all processing happens natively, the suggestions are quick and privacy is not jeopardized.
Interestingly, Google is not locking the AI Core to its own silicon. The company says it can take advantage of the machine learning engines built on silicon offered by Qualcomm, MediaTek, and Samsung.
That makes a lot of sense, as both MediaTek and Qualcomm now offer top-end, as well as mid-tier processors that support AI models worth billions of parameters. Moreover, the inherent architecture leaves the doors open for developers to fine-tune it for their own apps and develop custom AI-powered features of their own.
In addition to AI features built atop the Gemini Nano model, Google is also improving the Photo Unblur feature to better recognize pets like cats and dogs. Taking a APGE out of the Magic Eraser book, the Pixel phones get a new Clean feature that can remove smudges and gunk visible in scanned documents.
The Google Pixel Fold gets a new Dual Screen Preview mode that lets you deploy the outer screen as a viewfinder. The ability to use the Pixel phone as a hi-res webcam is also rolling out widely for users.
Following in the footsteps of Samsung, Google is also bringing a Repair Mode to Pixels that locks all the user data in a secure environment while the device is being fixed. the Pixel 6, and all phones launched after it, will also now show contextual quick replies when the Google Assistant is screening calls, offering users a quick way to answer calls with a text without ever picking it up.
Thanks to Call Screen, Google is also bringing contextual incoming call alerts to the Pixel Watch. Just like the iPhone-Apple Watch camaraderie, the Pixel Watch can be used to unlock Pixel phones. The first-generation Pixel Watch, following the December feature drop, is now able to sync Do Not Disturb and Bedtime Modes with your Pixel smartphone.