A Cambrian Explosion In AI Is Coming by Dag Kittlaus.
From the post:
However, done properly, this emerging conversational paradigm enables a new fluidity for achieving tasks in the digital realm. Such an interface requires no user manual, makes short work of complex tasks via simple conversational commands and, once it gets to know you, makes obsolete many of the most tedious aspects of using the apps, sites and services of today. What if you didn’t have to: register and form-fill; continuously express your preferences; navigate new interfaces with every new app; and the biggest one of them all, discover and navigate each single-purpose app or service at a time?
Let me repeat the last one.
When you can use AI as a conduit, as an orchestrating mechanism to the world of information and services, you find yourself in a place where services don’t need to be discovered by an app store or search engine. It’s a new space where users will no longer be required to navigate each individual application or service to find and do what they want. Rather they move effortlessly from one need to the next with thousands of services competing and cooperating to accomplish their desires and tasks simply by expressing their desires. Just by asking.
Need a babysitter tomorrow night in a jam? Just ask your assistant to find one and it will immediately present you with a near complete set of personalized options: it already knows where you live, knows how many kids you have and their ages, knows which of the babysitting services has the highest reputation and which ones cover your geographic area. You didn’t need to search and discover a babysitting app, download it, register for it, enter your location and dates you are requesting and so on.
Dag uses the time worn acronym AI (artificial intelligence), which covers any number of intellectual sins. For the scenarios that Dag describes, I propose a new acronym, UsI (user intelligence).
Take the babysitter example to make UsI concrete. The assistant has captured your current (it could change over time) identification of “babysitter” and uses that to find information with that identification. Otherwise searching for “babysitter” would return both useful and useless results, much like contemporary search engines today.
It is the capturing of your subject identifications, to use topic map language, that enables an assistant to “understand” the world as you do. Perhaps the reverse of “personalization” where an application attempts to guess your preferences for marketing purposes, this is “individualization” where the assistant becomes more like you and knows the usually unspoken facts that underlie your requests.
If I say, “check utility bill,” my assistant will already “know” that I mean for Covington, Georgia, not any of the other places I have resided and implicitly I mean the current (unpaid) bill.
The easier and faster it is for an assistant to capture UsI, the faster and more seamless it will become for users.
Specifying and inspecting properties that underlie identifications will play an important role in fueling a useful Cambrian explosion in UsI.
Who wants a “babysitter” using your definition? Could have quite unexpected (to me) results. http://www.imdb.com/title/tt0796302/ (Be mindful of your corporate policies on what you can or can’t view at work.)
PS: Did I mention topic maps as collections of properties for identifications?
I first saw this in a tweet by Subject-centric.