By Paul Miller December 10, 2014

AI for wearables

When software gets personal

Last week, Evernote CEO Phil Libin explained a really important distinction in types of AI-aided computing. The three approaches, as he sees them, are as follows: something that acts like a benevolent parent and treats you like a baby, something that acts like an assistant and makes you into a powerful executive, and something that gives you superpowers. He thinks of his new product, Context, as something in the third category. Context looks at your notes as you type them, and offers relevant links and notes. It helps you work smarter, not by interrupting and guiding you, or by requiring you to ask for assistance. It's just a brain add-on, not a wizard or a servant.

Libin's categories highlight the shortcomings of AI I explored in the first part of this series. AI is currently too dumb to know you, and therefore too dumb to know what's good for you like a parent or follow your instructions as effectively as an assistant. But software can be good at many things human brains aren't perfect at, like math, memory, and search.

Using software for the right purpose, and no other purpose, becomes absolutely important in the world of wearables. Put simply, wearable devices are more personal than any computer-like device has ever been, and so types of interaction and function that are appropriate for a laptop or a phone seem wrong or even intrusive on your wrist or embedded in a pair of glasses.

A couple of years ago I wrote about the history of wearables, which had a heyday in the 90s at MIT's Media Lab. What became clear to me is that early innovators in this space understood this distinction Libin is talking about, and wrote really great philosophy on the topic.

Steve Mann, often referred to as the "father of of wearable computing," was their foremost philosopher, and wrote this definition of computer augmentation:

"Traditional computing paradigms are based on the notion that computing is the primary task. Wearable computing, however, is based on the idea that computing is NOT the primary task."

So, what does this have to do with that Fitbit you just bought at Best Buy? Basically, it means that if seen as just another computer, a Fitbit uses accelerometers and algorithms to count how many steps you take, then syncs to a mobile app to display that information to you. But as a human augmentation device, a Fitbit is something that inspires you to walk more, and therefore become more physically fit. In fact, a Fitbit might serve its highest purpose if it trains you to be so physically active that you no longer need a Fitbit for motivation.

In a recent (possibly satirical) essay for The New Yorker, David Sedaris details his obsessive relationship with a Fitbit:

"Before the Fitbit, once we’d eaten dinner I was in for the evening. Now, though, as soon as I’m finished with the dishes I walk to the pub and back, a distance of 3,895 steps."

Before long, Sedaris spends most waking hours walking, leading to sore feet, lost weight, and an increased knowledge of the surrounding countryside. For Sedaris, the Fitbit wasn't functioning as a computer, it was functioning as willpower augmentation.

Until off-the-shelf wearable devices can give us more literal superpowers, like x-ray vision or super strength, their primary influence is on our attention. A Fitbit, on the simple end of the spectrum, reminds you of how (in)active you've been that day. Google Glass or the Apple Watch, on the other hand, have a limitless range of topics they can inject into your consciousness. Text messages, emails, Twitter, and New York Times headlines can all be whitelisted for brain interruption.

For Sedaris, the Fitbit wasn't functioning as a computer, it was functioning as willpower augmentation.

In contrast, earlier research on wearables offered more personal and contextual interruptions: the name and bio of the person you're talking to, a warning that you seem to be stressing out, a reminder to grab the milk on your list before you leave the grocery store. Something more like AI than apps.

I'm certain we'll get there eventually, and it's likely app-driven devices like Apple Watch and Google Glass will be the first to unveil these more interesting and advanced uses of wearables. But, ultimately, my hope is that these functions will then make their way back into more "invisible" devices like the Fitbit. I want to be augmented, not distracted.


About Paul Miller

That guy who left the internet for a year