You may not realise it immediately, but a lot of your phone’s functionalities are actually fuelled by AI. The technology is running on your phone behind the scenes even when you are not using it, handling a variety of tasks at all times. It analyses your phone’s usage to squeeze more power out of the battery, helps you take crisp photos, identify music, assists you in understanding a different language, and a lot more.
Until some time ago, AI was reserved for expensive equipment that used the most advanced technology. But now AI has become such an integral part of mobile applications that chipmakers felt the need to develop AI chips dedicated to machine learning and deep learning tasks for faster processing. Here’s a look at the various ways AI is being used in phones so far.
AI in smartphone voice assistants
Google Assistant, Siri, and Bixby are the most popular voice assistants right now, and you can access at least one of them on any smartphone. Even the most basic of functions of these voice assistants involve the use of AI. The “Hey Siri” detector, for example, uses a Deep Neural Network (DNN) to convert the acoustic pattern of your voice into a “probability distribution over speech sounds,” according to a research blog post by Apple. It then uses a process to computer confidence score to ascertain if the phrase you uttered was “Hey Siri.”
But that’s only the tip of the iceberg. Siri, Google Assistant, and Bixby are built to process and understand much more than prompts like “Hey Google” and “Hey Siri.” They instantly process complex queries and respond by constructing clear sentences as a human would. These sentences aren’t saved responses – the AI assistants are trained with the rules of linguistics to help them construct sentences, similar to chatbots like ChatGPT.
The use of AI and machine learning is crucial in voice assistants because of the sheer amount of things they can help you with – set reminders, give directions, read out news, control your devices, send messages, and a lot more. To program these tasks manually would have been a gargantuan task, so voice assistants are trained to learn from experience and make decisions on their own.
AI in smartphone photography
While AI image processing was initially limited to high-end smartphones, even entry-level offerings include them now. AI in imaging ensures that your snap looks presentable no matter what the ambience or your skills are. It works its magic by enhancing images, sharpening blurry elements, reducing noise for pictures taken in low-light scenarios, adding a bokeh effect, and a lot more. AI is especially handy in budget smartphone cameras because it help make up for the lacking in hardware with software processing, significantly improving quality.
A good example of AI application in smartphone photography would be the True Tone tech featured on Google Pixels. Camera technology has historically failed to accurately capture people of colour resulting in unflattering photos for those with darker skin tones. With True Tone, Google improved its camera tuning models and algorithms to highlight diverse skin tones more accurately.
Another example is Samsung’s Single Take feature. Single Take ensures that “precious moments that pass in the blink of an eye are not lost,” according to Samsung. It records video for up to 20 seconds while simultaneously snapping up to 10 pictures per second. Once that’s done, AI adds various effects to highlight the best photos and videos, generating up to 10 photos and up to 4 video clips in real-time.
AI in smartphone face recognition
Android smartphones do support facial recognition but they push users to use the far more accurate fingerprint tech that just about every phone running the OS comes with. iOS is where the real facial recognition action lies. First launched on the iPhone X in 2017, the tech replaced Apple’s Touch ID fingerprint scanning system. It’s now a staple across all the latest iPhones. It works using a number of hardware components such as the TrueDepth camera system and the AI-supporting Bionic chips.
The AI kicks in when the iPhone checks scans of your face against the one you’ve set up and stored on your device to see if it’s a match. AI helps Face ID to adapt to changes in your appearance such as wearing makeup or growing facial hair. Over the years Apple has updated Face ID tech, increasing its accuracy and with the iOS 14.5 update, Face ID even works when you are wearing a mask.
AI in augmented reality applications
Dedicated smartphone chips boast cutting-edge AI processors, known as Neural Processing Engines or similar, that power on-device Augmented Reality experiences like Apple’s iconic Animoji. As VR gains traction and its possibilities unfold, AI continues to showcase its abilities in real-time object tracking and recognition. Scene recognition through the camera lens is a great example of AI’s practical applications. Devices now have the capability to distinguish between various types of scenes, such as animals, scenes like sunsets, and text-filled images.