Apple is looking to increase its knowledge of artificial intelligence, by hiring more experts in machine learning. It has posted dozens of job listings relating to artificial intelligence and has reportedly started to recruit from PhD programs in the field, with the ultimate aim likely to involve making
Siri more intelligent, to the point where it can offer information to the user before they even need to search or request it in the first place.
Reuters reports Apple is seeking at least 86 more employees involved in machine learning, and it has apparently pushed hard in the area, with some academics claiming it is now stepping up its efforts to pull in academic researchers. While Apple does not say how many of its employees work in machine learning, one former employee suggests the number has tripled or quadrupled in the last few years alone. The job listings in question relate mostly to software and services such as Siri and iOS search, though machine learning experts for product marketing and retail are also being sought.
While Google does have the advantage, with regards to Google Now's cards, Apple has already started to claw back its in-advance intelligent assistance. In iOS 9, Siri will become more proactive, suggesting apps to the user based on their usage habits, automatic identification of unknown callers through searching correspondence, and the generation of appointments and reminders from a received email automatically.
List of machine learning job openings at Apple
Despite the hiring of more staff, Apple may be finding it harder to find employees than competitors, such as
Google, due to its privacy policies. Experts believe the standards Apple holds for protecting user data makes it difficult to analyze information created by its users. For example, while data related to Siri an be held and monitored for up to six months, services such as Apple Maps only have 15 minutes to look at the data. These small windows are said to be unattractive to data scientists and machine learning experts, due to the need to access large quantities of data.
Apple's policies may also harm the way the company processes data for machine learning applications. Since Apple prefers not to keep sending data back to its servers if it is not required for the task, in order to protect the privacy of the user, more processing will have to be performed on iPhones and iPads rather than allowing a cloud server to do the same thing, something which also means the process has to be accomplished with little access to the data of other users for comparison.
These constraints may still prove attractive to some machine learning experts, suggests assistant professor at Stanford University John Duchi. Matching Google's own work while having less readily available access to the required data could be seen as an interesting problem, with the possibility of forcing Google to reconsider its own policies if Apple is able to keep up despite the restrictions.