It's no secret that Apple (AAPL -0.89%) has spent a great deal of time, money, and resources to make its devices -- particularly the iPhone -- smarter using artificial intelligence (AI). One of the factors that differentiates Apple from the other companies developing the technology is its consistent commitment to data privacy. To advance that agenda, Apple has focused on on-device AI rather than taking a more cloud-centric approach, which helps explain the company's latest acquisition.
Apple has reportedly acquired AI start-up Xnor.ai for about $200 million, according to a report from GeekWire, citing "sources with knowledge of the deal." Xnor specializes in developing complex AI systems that are extremely efficient, able to use minimal amounts of power and run locally on devices like smartphones rather than by transmitting information to remote data centers. The company has also created algorithms that specialize in capturing and processing images on small devices, a computing-intensive task that has historically been accomplished in the cloud.
Cutting-edge AI tech
The technology developed by Xnor has already been used in a variety of applications. It was integrated by smart camera company Wyze in its "person detection" feature before being unceremoniously removed recently "due to the unexpected termination of our agreement with our AI provider," undoubtedly the result of Xnor being acquired by Apple. The company is also developing an application that would help improve inventory management by monitoring shelves in retail stores. Previous advances include the development of an AI chip that can run for years using a tiny battery or even a small solar-powered cell.
The technology Apple is gaining could be used to improve future versions of the iPhone by augmenting its existing camera capabilities, or it could be used by developers to create apps as part of the company's Core ML (machine learning) toolkit. It also continues a multiyear push by Apple to run AI on its devices as a way to help ensure the privacy of its users.
The cloud versus the edge
Deep learning, a branch of AI research, involves amassing huge amounts of data and collecting it in a central location in the cloud for processing. Historically, this has been necessary because the process is computationally intensive and requires complex calculations, and the sheer volume of data can be enormous. Apple's stance on data privacy is well documented, one the company wasn't willing to forego in order to keep up with its big tech rivals in the race for AI.
As a result, Apple has taken a different path. Over the years, the company has focused more on developing the next generation of AI systems that rely more on using minimal information and processing the data on the device itself, called at-the-edge or simply edge computing, rather than sending user information to the cloud.
It's still necessary for Apple to process some of its data in the cloud, but the company obscures personally identifiable information using a process known as differential privacy. In order to protect user information, the technology injects "noise" or random data into the information it collects before sending it to Apple's servers, which helps anonymize the data and protect the user's identity.
Not the first time
This isn't the first such small AI-focused company Apple has acquired. In 2016, the company acquired AI start-up Turi, a company that created AI-based tools for app developers that could scale to a large number of users. This became an integral part of Apple's Core ML app developer frameworks.
Apple will no doubt continue to acquire smaller AI companies to augment its internal AI efforts, particularly when it comes to the area of data privacy, and we'll no doubt see the integration of this technology into Apple's products in the coming months and years.