And now it helps to push the limits of helpfulness in a smartphone. With machine-learning concurrency at its core, the Tensor processor enables high-quality video, improved security, and new user experiences that make Pixel more personal and helpful than ever before. And it allows Pixel to adapt to you, accommodating the different ways we use our phones.

But this secret ingredient of Pixel phones – what makes them more powerful and helpful¹ – didn’t happen overnight. 

A few years ago, Google’s researchers came together across hardware, software, and machine learning to build the best mobile computer chip. It had to realise a vision of what should be possible on Pixel smartphones. The result was Google Tensor.

While all advanced smartphones come equipped with powerful processors, the Google Tensor chip stands out because it is focused on delivering complex artificial intelligence and machine learning processes on the phone itself instead of sending data to the cloud, essentially bringing cloud computing to your smartphone. And because these algorithms run on-device rather than through a network and server, they work fast.²

More good news: Tensor keeps getting better. Pixel 8 and Pixel 8 Pro are equipped with the latest model, Google Tensor G3. Not only have the number of machine learning models on device increased by 400%, but the complexity and sophistication has increased as well. The largest machine learning model on Pixel 8 is 130 times more compute heavy compared to the same on Pixel 5. 

As a result, Google AI and Pixel 8 give you even more of the cutting-edge photo and video features that Pixel is known for, and new, smarter ways to get things done throughout your day.

These AI and ML capabilities are becoming “more and more prominent and integrated into our daily lives,” says Nesra Yannier, who researches human-computer interactions at Carnegie Mellon University. As people become used to the benefits of personalised, intuitive technologies, she adds, “it’s becoming something they expect and feel is natural in technology products.” 

Tensor SoC (System-on-Chip) is the processor developed by the Google Research Team. It was conceptualised in 2016 and first integrated on Pixel 6 in 2021. Now Pixel phones use an AI processor and machine learning powered by Tensor to perform increasingly advanced tasks, like instantly translating messages and videos³ without internet.²

And Pixel phones get better with time, becoming more personalised and helpful, ⁴turning them from a one-size-fits-all piece of hardware into a device that’s more helpful for you. Pixel can distinguish a tree from a person in a photo, or understand if the picture is blurry and in need of a touch-up. 

“Your phone needs to understand your world – your context, your unique needs and preferences, how you speak, and what you care about,” said Rick Osterloh, SVP of Google Devices and Services. With Tensor bringing Google AI to Pixel, the phones go a long way toward delivering on that promise:

• Take better photos and videos – The most visible examples of machine learning in action on your phone can be found in your photos and videos. With Tensor G3’s cutting-edge computational photography and image processing, your phone can now render more dynamic images, capture more details during zoom, and process sharper, higher-quality photos and videos. And it can do this approximately half-second faster than the previous generation.⁵ Pixel phones can detect faces in a shot and adjust the focus and brightness to perfect the exposure. With Real Tone, photos represent the nuances of different skin tones beautifully and authentically. If someone moves, Pixel can fuse multiple shots to remove the blur with Photo Unblur.⁶ Night Sight combines multiple shots into one frame to optimise for low-light scenery. Magic Eraser on Pixel 8 can remove larger distractions by better predicting backgrounds, creating higher-quality photos.⁷ And let’s not forget Best Take: Get the perfect group photo, every time, by fixing closed eyes, misdirected gaze, facial expressions, and more. Best Take uses Tensor G3 to merge multiple images and help everyone look their best.

• Have better phone calls – With the help of Tensor G3, Pixel’s Call Screen can now detect and filter out even more spam calls. ⁸ And improved Clear Calling gives you high-definition, crystal clear phone calls.⁹

• Communicate faster – Machine learning recognises speech and turns it into text through Live Caption, in real time. Google Tensor also makes real-time language translation possible with Live Translate, either speech-to-text or directly to audio.¹⁰

• Use your phone longer – Google Tensor runs more advanced, state-of-the-art ML models but at lower power consumption compared to previous Pixel phones.⁴ This helps save power so your battery lasts longer.

• Protect your data – There’s a separate core on the Google Tensor chip that’s set apart from the application processor, so sensitive tasks and controls run in an isolated environment, making them even more resilient to attacks. The Tensor Security core works with the Titan M2™ security chip, adding another layer to better protect your personal information like texts, photos, sensitive data and more.¹¹ And Pixel 8 uses advanced machine learning models to deliver a more secure Face Unlock that meets Android’s highest biometric standard class, making it strong enough for banking applications.

• Get more relevant suggestions – Google Tensor helps Pixel surface the content you need, when you need it. Simply glance at Pixel’s always-on display to find out when you need to leave for your next appointment, or see what song is playing with Now Playing song detection.

• Get help when you need it – If Pixel senses that you’ve been in a car accident, car crash detection can help reach 999 when you can’t.¹²

• Pixel uses your device’s location, motion sensors, and ambient audio to detect a possible car crash and contact emergency services if you don’t respond to a prompt.¹²

• Take better pictures, even if you’re visually impaired – The helpful Guided Frame feature is better with Tensor G3, too, moving beyond selfie assistance for blind and low-vision users. Guided Frame now works on both the front and rear camera, and now it recognizes elements like pets, food, and people, not just faces. It’s just one of the many helpful features built to make Pixel phones more accessible for more people.

Pixel phones learn the different ways we use our phones so they can provide the right information at the right time. 

For instance, understanding the nuances of human speech is a challenge that’s unique to each person. It means understanding syntax, intent, and the context of your request, and knowing the terms and names that are important to you but potentially uncommon to others. It also can understand your accent and dialect, isolating your voice when there’s background noise, and even hearing you correctly when your mouth is full.

“If these products are invisible and user-friendly and natural, they will become part of the everyday experience,” Yannier says. “That’s how technology should really be.”