Some time ago, experts from the tech industry agreed that Apple has a cap on its artificial intelligence (AI) program, and it was mainly due to the company's high privacy factor.
Recently, Apple executives talked to Backchannel in great detail about the development of AI that takes place inside iOS. Most users might not even acknowledge how much AI is actually present in key features of iOS, and it might be that Apple wants it this way.
According to senior vice president of software engineering Craig Federighi and senior worldwide marketing vice president Phil Schiller, AI is far from being the "final frontier." The executives noted that rival companies make AI sound like the final goal of research, but for Apple this is just a regular task in a day's schedule.
Backchannel's Steven Levy talked to the Apple execs and took a peek at a two-page agenda where all the machine learning-imbued Apple products and services were displayed. Some are already embedded in the current products of the company, while others will land soon.
To give a few examples: Siri went to using neural networks to enhance its voice recognition abilities in 2014, after years of users complaining about the virtual assistant's mistakes. The modification has sharply changed the way in which Siri now behaves, and fewer mishaps have been reported.
Owners of iPhones also make use of machine learning when their phone suggests the name of a caller who does not appear in the contacts list but who recently sent them an email. By studying user patterns, iPhones are ready to predict a group of apps you might prefer using.
Another nifty machine learning trick is that the hotel where your reservations are made shows up in the search bar in Maps. Also, the software that enables the Apple Pencil to feel so natural was built using machine learning.
Apple taps into deep learning tools to find and uncover fraud on the Apple Store, to help users get more battery life out of their devices and to make it easy to pick up helpful feedback from the myriad of reports that arrive via beta testers.
"Our devices are getting so much smarter at a quicker rate, [...] which enables more and more machine learning techniques," Schiller said.
At the same time, the execs confirmed what most techies assumed: Apple does not, in fact, have a dedicated AI or machine learning department. However, Apple is constantly recruiting talent for its AI development. Only this August, Apple bought Turi, an AI startup that should help the tech company expand its machine learning know-how.
Some find that Apple's limitations are twofold. For one thing, Apple's AI problem sprouts from the lack of a search engine, which would be able to output data to build complex neural networks. For another, the die-hard loyalty to protecting user information blocks Apple from using some data to help it advance its machine learning processes.
We are confident that the company will find creative ways to merge intelligence and machine learning for the predictive features of the future, but a question remains: how much evolution can be expected from such features if they are neutered by Apple's privacy limits?
