Interaction is a natural outgrowth of increased data and more elaborate presentation. The more practical questions relate to when will interaction with a digital persona begin, i.e., how much data, how elaborate a presentation?
        Moreover, not only when, but also how real will these digital personas be? Will they be able to pass the famous (in computer circles, anyway) Turing Test? Essentially, the Test asks whether human judges can determine, by typing questions, if they're interacting with a computer program generating responses or a human typing those responses.
        In actuality, creating a realistic digital persona that can fool the humans is easier than creating the software that can effectively recognize human speech. After decades of development, speech recognition is at best limited and crude, relying on the human to verify, and correct, the text generated by the speech. Not very effective for a realistic interactive conversation with a digital persona.
        Of course, the Web is already interactive: you click on This and you get That. In addition, digital personas need only a modicum of interaction before we're likely to see them sought for relationships by the same kind of people who now seek relationships with Internet strangers and prisoners.
        At least that's my opinion. One thing we do know: the more realistic the presentation, the more people will expect an enhanced interaction. How enhanced? That depends on the same two things it's always depended upon: hardware and software. And one other: bandwidth. Cost is irrelevant. Someone will be able to afford whatever can be produced. And the rest of us? Well, there's always books. I hope.