Machine learning, artificial intelligence, and deep neural networks are all red-hot topics right now. After decades of relative obscurity, these technologies are viewed as the key to driving the next generation of computing advances. Apple has been viewed as playing catch-up in these fields in some respects and the company is reorganizing its efforts to centralize them. It’s also hired John Giannandrea, who used to head Google’s search and AI team, to oversee the new division.
Giannandrea will head a new organization that includes Siri, Apple’s Core ML team (that’s the machine learning API that Apple launched to make AI and AI-focused tasks run more efficiently on iOS devices), The Verge reports, as well as Apple’s machine learning division as a whole.
Apple is widely seen as lagging on the AI/ML market when compared with its competitors like Amazon and Google. The company’s Siri-based HomePod was a rare misfire for the company — a speaker whose principal goal seems to have been to lock you into using Apple (and only Apple) services and to ruin countertops. Apple has said nothing about HomePod sales, but they aren’t believed to be very good and Apple is said to have captured 4-6 percent of the smart speaker market after it launched the platform.
I have no use for digital personal assistants in any form, from any company, but I do periodically test Siri to see if the service has improved. It hasn’t. The traditional claim is that one reason Siri is going to lag behind is because Apple doesn’t capture the absolute oceans of data that Google, Facebook, and other Silicon Valley companies have been hoovering out of our lives for a decade or more now. This may well be true, but I suspect it doesn’t tell the entire story. Amazon, for example, continues to work on its own AI projects — and Amazon prominently split with Google to the point of developing its own version of Android because of its concerns over Google’s privacy policies and data use.
Furthermore, it’s not clear why Apple tapped Giannandrea, specifically, to lead this charge. Apple has made a great deal of noise about its commitment to user privacy, and has used this to rhetorically differentiate itself from companies like Google. Bringing in someone who oversaw AI and ML development at Google using the same data sets that Apple doesn’t have and supposedly doesn’t gather doesn’t make a lot of sense, unless Apple plans to make some major changes to its data gathering practices at the same time. Nothing along these lines has been announced.
You can read this situation a couple different ways. On the one hand, some will conclude that Apple either performs the same types of data collection that Google does or that it wants to, and that bringing Giannandrea onboard is an effort to match the search giant in this area. But another potential option is that the value of Google’s data about its users may be less central to the question of why Siri has lagged behind its competitors than is immediately apparent. We’re not claiming the data hypothesis isn’t true — we’ve got no insight on it one way or the other — but it’s also the sort of explanation that could be a contributing factor without constituting the overriding reason why a situation looks the way it does. Either way, Apple is clearly planning to make some significant changes to its AI/ML development programs. How they evolve should tell us useful things about what’s been holding Apple back.
Google’s AutoML Creates Machine Learning Models Without Programming Experience
The gist of Cloud AutoML is that almost anyone can bring a catalog of images, import tags for the images, and create a functional machine learning model based on that.
Google’s Cloud TPU Matches Volta in Machine Learning at Much Lower Prices
Google and Nvidia both offer competitive machine learning products — but Google is beating Nvidia on costs, at least in certain tests.
Intel Positions Xeon as Machine Learning Competitor in Inference Workloads
Intel is tossing its own hat into the machine learning ring, positioning Xeon as a potent inference workhorse.