Based on the recent report by Gartner on artificial intelligence, the site stated that the technology will be included in almost every new software product by 2020.
With the rise in demand for high-tech processing in mobile devices, Google is looking into enhancing its app services with the help of artificial intelligence via their proprietary innovation firm Alphabet.
The search engine company is working on making things easier and more efficient for businesses by employing AI in Android apps, which in turn will make their platform more appealing than Apple’s iOS mobile operating system.
Google predicts that Android device makers will come out with new mobile devices that feature digital signal processor chips that support methods for deep learning. This is the type of artificial intelligence that the search engine and other technology companies have been using inside their apps over the past few years.
As a forward-thinking global brand, Samsung has included deep learning in their propriety built-in mobile AI, Bixby. The virtual assistant that arrived along with their current flagship device is able to learn the user’s mobile routine based on their usage and provide them with the right content when they need it. On the Galaxy S8 page featured by O2, it states Samsung’s AI promises users to make things easy for them to do what they want, when they want it by recommending content and information based on their content consumption, internet usage, overall mobile habits and behaviors.
Author Kevin Tran wrote in his BGR post about Apple’s AI that the California-based tech company is now also developing its own dedicated chip and software that runs artificial intelligence and performs deep learning. Internally known as the ‘Apple Neural Engine,’ the chipset will manage artificial intelligence processes, including biometric sensing, facial recognition and speech detection. The chip will eventually be integrated into the iPhone and iPad hardware, but no concrete date has been provided regarding its release. The company’s CEO Tim Cook has previously stated that AI will be dominant in the smartphones of the future.
Historically, the process of deep learning was executed by powerful computers in a secure data center, where the results were later fed back to the mobile devices. However, Google projects that there will come a time when smartphones and tablets will be able to undertake deep learning processes themselves.
The deep learning process in technology involves two steps. Initially, researchers train software in various data, from images, videos, and patterns. Then, they present new data to the neural networks and command them to make inferences. This step allows the computer to identify images based on its associated word, making it easy to provide the most accurate answers and solutions based on request.
While it sounds easy to train computers, it is not always the case as AI comes with its own limitations just like any technology today. There are some innate human skills that computers are not able to replicate, such as human emotions, negotiating skills and more.
In July, Facebook killed its AI program as researchers found out that the technology lacks “complex communication and reasoning skills, which are attributes not inherently found in computers.” The team also found out that the technology started talking in coded language after they tried switching English to another language. It has been reported that Facebook, regardless of the setback, continues to work on producing a smart AI for their apps.
Once rectified by Google, users can expect smarter and more helpful apps that can process requests faster without having to call back to data centers. Built-in AI in apps could refine systems that require image or speech recognition quickly. In a broader sense, mobile devices will also have more computing power onboard.
Dave Burke, VP of engineering for Android at Google, announced at the Google I/O last May that the company will make use of that added computing power by launching a new application programming interface for neural networks. AI will be part of the TensorFlow Lite, a special version of Google’s open-source software for deep learning. Burke said: “[TensorFlow Lite is] a library for apps designed to be fast and small, yet still enabling state-of-the-art techniques." Last year, the company improved the software by announcing its iOS support.
Google’s announcement came after Facebook launched their open-source deep learning network Caffe2 that has a support for Android and iOS apps with source code available on GitHub. The system offers developers greater flexibility for creating high-performance products efficiently. Similar to Google, Facebook has also built partnerships at the hardware, device, and cloud levels, targeting market leaders in each category.
Meanwhile, Apple is still yet to add tools for incorporating artificial intelligence systems into their iOS apps. The company is also working on two of their own deep learning frameworks: Basic Neural Network Subroutines (BNNS) and Metal Performance Shaders (MPSCNN). However, these two frameworks have been found to be much less capable than others, including TensorFlow, with its limited deep learning capabilities.