Apple Takes Predictive Analytics Mainstream
The biggest tech company in the World is about to teach us we may be more predictable than we realized.
Yesterday, Apple released iOS9, which is either a life-changing or inconsequential OS upgrade, depending on which tech blogger you ask.
Its most important upgrade is set to become the most useful feature you won’t notice. Apple has brought predictive analytics to your handset, and you probably aren’t going to like what you learn about yourself.
While many of the core elements of the new predictive features on iOS9 have been available on Google Now for some time, Google required the usage of its core services to mine your data. If your behavior wasn’t noticeable in your Gmail inbox or your browsing activity in Chrome, it wasn’t baked into the Google Now predictive results you received. It made sense: as they say, if a product is free, you are the product. And Google was already building predictive analytics into its ad services, targeting you better and more aggressively when you’re in purchasing mode. Google Now just made some of their data useful to you, which turned out to also be a tool to keep you using their core services.
That being said, Apple’s very different business model has allowed them to open up predictive analytics to everything on your device, all the while keeping your data private. After all, Apple wants to sell you phones more than they want to sell you to advertisers.
Apple does see our collective data in aggregate, but that data on user activity simply proves something sociologists have been trying to tell us for decades: we are incredibly predictable.
Our common behavior is becoming easier and easier to mine and understand, primarily because we are a species that lives in patterns. Apple knows when you start an email to Eliza in purchasing, you usually include DeMarcus in accounting. Apple knows when you get an email confirming your dinner reservation, you probably want to put it on your calendar as well. And Apple knows when that reservation rolls around, if you aren’t already on the road and traffic is looking rough, you’d appreciate a heads up so you make it in time.
The term “predictive” is noticeably missing from Apple’s description of its new features. Apple calls it “proactive.” This is further evidence of Apple’s incredible ability for marketing and positioning.
It’s a great turn-of-phrase. Why? Because proactive feels personal, whereas predictive feels mechanical and numerical.
But make no mistake, Apple’s “proactive” assistant is simply one of the most customer-facing predictive modeling tools in the market today. The question is, will we notice?
There’s good reason to think this technology will sink into the background, becoming an expected part of any digital experience, but Apple will have to walk a fine line between helpful and creepy in order to not draw too much attention to its data collection skills.
If, for example, you tend to casually take a break from working most days sometime between 2:00 and 3:00pm, browsing Instagram, Apple’s algorithm would be taking it too far to proactively open the app when you unlock your phone at 2:13pm on a Thursday. You may not have even noticed you do that and if Apple picked up on it you might be reasonably freaked out, despite the fact that the data on that activity is just as clear as your propensity to listen to a podcast when you plug your headphones in.
Likewise, predictive analytics can accidentally reveal your darkest secrets simply because to the algorithm, all data is equal. If, you typically send photos on your phone to either your husband, Mike, or your illicit boyfriend, who you think you’ve cleverly listed as “Nail Salon” in your contacts, Mike might just realize what’s up when you go to send a selfie of the two of you to his mom while hanging out at the Orioles game and “Nail Salon” pops up as a contact you might want to text that pic to.
It’s interesting how we spend decades solving massive technological problems like predictive analytics, and realize their implementation starts to present us with incredibly human problems as a result. This is the major challenge of the data revolution. What is enough and what is too much when it comes to personal data mining and predictive modeling? How do we make our algorithms behave with tact and prudence?
As it turns out, Apple has just placed itself at the center of that challenge… and you are the experiment.