There is reason to believe that over the next few years (or at the very least, decades), we will develop computer-based, general purpose, artificial intelligence -- what some people are calling artificial general intelligence or AGI. The implications of this will be far reaching, but as this isn't a book, I'll concentrate on the impact on computer UI.
Anyone who spends time with children will realize that they follow a natural progession in how they communicate information they've previously acquired: first they repeat it, then they retell it in their own words, then they retell it from their own perspective, and finally they retell it from the listener's perspective. As each of these stages is more difficult than the preceding, there is reason to believe this progression is inherent, and will be followed by AGI.
Today, we are firmly in the repeat it phase. Ask an Internet search engine for information about, say, electric cars, and it will return pointers to a huge number of sites containing information on or related to the subject. While the mechanism behind this feat is complex, it is not intelligent in any real way. Retelling the information in different words is the first sign of a true AGI being involved, as it requires an understanding of the material at something above the syntactic level. Retelling the information from a different perspective is one step further, but isn't terribly interesting from a UI perspective, beyond its use by politicians and corporate communications departments.
When things really get interesting is when the AGI develops what's known as a "theory of mind" -- the ability to reformulate things from your perspective rather than its own or some arbitrary one. At this stage, it can leave out things you already know. It can use metaphors you can rapidly grasp since they relate to areas with which you're already familiar. And it can relate the information to other items you're known to have an interest in.
(Note: The lack of a theory of mind is one of the aspects usually associated with autism. It's worth considering how much computers have in common with sever autism -- lack of social skills, inward focus, single-mindedness, etc.)
Of course, it can only reframe things in your perspective if it knows what your perspective is. Knowing that requires either watching you or working with you. Most likely, the process will start with watching, initially the easy things to watch like your online activities -- web browsing, email, calendar -- and later the harder things like where you go and who you talk to and what you say to them (though a GPS phone and a little speech recognition software could cover a lot of that ground). But as AGI develops, it will be more and more a collaborator rather than a tool.
It may eventually understand you so well that it knows what you will want before you know it yourself. At which point, you won't have to even ask a question, just listen to the answer.
Comments