Genuine artificial intelligence and inference may be currently beyond our grasps, but hand-programmed special cased procedures (what you might consider the prod

Genuine artificial intelligence and inference may be currently beyond our grasps, but hand-programmed special cased procedures (what you might consider the product of reasoning, rather than the existence of) are certainly doable. If the program behind Siri can determine that you want to know the location of your next meeting, and sift through your calendar event data for the next meeting in it, the human engineer behind this scenario should anticipate that the user is likely to want directions to their destination as well, and make the program offer them while retrieving them in the background. It's also certainly doable for the program to learn, with simple machine learning techniques, which addresses the user already knows how to get to (and thus does not want to be offered directions to), with a variety of data sources from "did they decline when offered directions?" to analyzing GPS data from previous trips to determine if they've been to that location before. I'd wager that an assistant service like Siri can implement a great deal of the everyday "functionality" of a human personal assistant with nothing more than precise interaction design, careful data logging, and machine learning models to tailor assistance to the individual needs of the user. That's not easy, especially considering each interaction has to be designed by hand, but it's more than possible with today's technology.