In order to pull this off your iPhone would need the necessary processing power, as well as a big enough offline knowledge database.
So far, Apple and other companies have invoked Machine Learning to make our lives better. But there’s also a dark side looming.
Apple has acquired Danish machine learning startup Spektral. Its technology separates people and objects from a background in photos.
Those who have an iPhone X have little incentive to upgrade to an iPhone XS. But the iPhone XS Max appears to be a winner. At least until pre-orders start for the iPhone XR.
Wired has a great article about how Apple made the A12 Bionic Chip. The A12 processor is even more powerful than its predecessor.
This scientist would be, “building transformative neurotechnology,” and dollars to donuts says it’s related to Apple’s ongoing neural net projects.
Mubert is an app that uses AI to generate focus music. By focus music I mean certain sounds that it claims will increase your productivity when you study, work, or do something creative. I don’t know of the science behind this, so it may or may not be a legitimate claim based on evidence. But I do know that sometimes I like to play certain songs as background music when I’m writing. I’ve tried Mubert out and the music it generates is nice to use in that way. The music is electronic, and it’s generated by taking certain patterns typical of electronic music and creating an endless stream of pattern variations. Mubert worked with techno producers to create and sample sounds based on different genres of electronica. App Store: Mubert – Free
Last year, YouPorn Foresights used AI to predict what the most popular search terms would be in porn. This year the company did something similar. The data science and machine learning teams trained a recurrent neural network to look at the current most popular performer names, and have now created what science has predicted that the next generation of stars will call themselves. There are 69 names, both male and female, and the results are hilarious. As you would expect from AI, the names sound weird and goofy. My favorite names from the list are Man Master, Al Gorr (obviously my future kid), Summer Sax, and Paris Buttomina. It’s a safe-for-work list that you can check out here.
It’s easy to ask Siri for directions to Trader Joe’s, Starbucks, and other big businesses. But what about small businesses?
I wrote about Kévin Eugène before when he created a macOS concept. Now he’s back with a Siri concept, and it looks great. The concept is called iOS Mogi, and it’s based on something called parallel help. Basically, it involves Siri being able to work in the background to carry out your commands, instead of the current “issue command, Siri reply, done, exit.” Siri opens as a small notification, instead of taking up the whole screen. This is the true future of virtual assistants. They should be able to do things on their own in a more proactive way than iOS 9 Proactive ever could.
You know how sometimes, you feel frustrated when you try to send a message with Siri, and end up taking your phone and typing your text? In iOS Mogi, instead of relying completely on Siri to do things for you, you can ask it to help you get things done faster. No more wandering in the UI, simply begin your sentence with « I want to… » and Siri will let you do it, without leaving what you were doing (in iOS Mogi, what you are doing is really precious).
Check out this video from OpenAI of a robot hand learning how to manipulate a block. This an incredibly difficult task, and the level of difficulty is one of the many reasons Apple needs humans assembling iPhones. OpenAI used machine learning and virtual simulations for the robot to spend 100 years of trial and error to learn what you’ll see in the video (TechnologyReview has more details). Those virtual lessons were then used by the real-world robot hand, and it’s pretty darned cool. Check it out.
Adam Christianson from the Maccast and Bryan Chaffin join Jeff Gamet to talk about Apple’s AI boss John Giannandrea, plus they share their thoughts on a leaked photo showing the next iPhone’s glass front.
Apple hired former Google AI chief back in April, John Giannandrea. Now he has been given the reins over Siri only three months after getting hired.
Microsoft is ramping up its stake in the artificial intelligence market by buying the AI and machine learning startup Bonsai.
Twitter has lost its corporate mind, Bryan Chaffin and Jeff Gamet argue in this episode of ACM. They also weigh the importance of WWDC 2018 in terms of Siri, and discuss whether or not Apple has to announce significant improvements to remain competitive in AI. Then there’s the revelation that the FBI exaggerated the number of locked iPhones it couldn’t get into, and they squeeze in a fourth topic, too: Apple’s hunt for a new campus, and how it contrasts with Amazon.
Every Friday the For You section is updated with what the algorithms think you’ll like. For a while Andrew Orr didn’t like his recommendations. But he thinks he discovered a way to improve Apple Music.
According to Apple, users would start off their Siri interactions with “Hey Siri” even when the only way to access the service was by using the iPhone’s Home button.
In a letter to employees, Tim Cook said that John Giannandrea shared Apple’s commitment to privacy and “our thoughtful approach” to machine learning.
A creative AI called SHELDON has created its own podcast, and the result is an infinite, personalized experience. SHELDON was created by James Ryan, a PhD student from University of California (named after Sheldon Klein, an early pioneer of expressive AI). The goal is to create a unique podcast experience for each user. When you listen to your first podcast episode, SHELDON randomly assigns you a county in the world it created. Each simulated country has its own characters with their own individual stories. On February 2 James released a proof-of-concept pilot version on Soundcloud, and he wants to release a beta version of the podcast in early 2019.
Apple usually avoids talking about its projects, but it also wants to reassure researchers that you can still publish papers as an employee.