Apple Inc. completely changed how Siri worked in the summer of 2014, and not many noticed it. Siri did not perform as advertised when it was first released on the iPhone 4S in 2011. The reviews were not good, but what was more problematic was that Siri’s set of feature was purposefully stunted by the Cupertino-based firm, says a report from BGR.
Apple Siri made a big leap forward
Siri was, in fact, not even as capable as the Siri app that was previously released on the App Store by its original developers, the report says. Ultimately, it became apparent that rival AI systems by Google were far better than what Apple was bringing to the tech industry. However, Siri started to improve slowly with time. Its speed recognition, and processing capabilities became better and the scope of its functionality expanded.
An article from Steven Levy – writing for Medium’s Backchannel – told how Apple’s ongoing research into machine learning made these improvements possible. Levy spent some time with two Siri scientists and top-level Apple Inc. executives – like Craig Federighi and Eddy Cue – to provide us with a glimpse into the incredible work the firm is doing with deep learning.
In an interview to Backchannel, Eddy Cue – Apple’s internet services boss – said “This was one of those things where the jump was so significant that you do the test again to make sure that somebody didn’t drop a decimal place.” As per Alex Acero, who spearheads the Siri speech team, Siri made half the errors using the new neural network than it had in the past.
All thanks to neural nets and deep learning
An iPhone user who has come across Apple’s AI will see that their iPhone identifies a caller who is not in the contact list but who did email recently. They also shortlist apps that a user is most likely to open next. You must have seen that a map location pops up for the hotel that you have reserved (even before you type in) and your iPhone points out the parking even when didn’t ask. These all techniques were made possible and greatly enhanced with help from neural nets and deep learning.
Another interesting fact from the article was how the tech giant leveraged its machine learning expertise while developing the Apple Pencil that first made its appearance with the iPad Pro. Reviewers praised the Pencil, noting that inadvertent touches via the palm were not paid heed to. It turns out machine learning was responsible for this as well.
As per Levy, Apple Inc. adapted voice recognition of Siri to use a cutting edge artificial intelligence technique called neural networks, and switched it over back in 2014 on July 30. Neural networks is actually a type of Artificial Intelligence inspired by the human brain, which has become very useful due to powerful computers.