I also posted this on O'reilly's site.
Edit: After thinking about it a bit, having the device as another "person" that everybody interacts with might be a way around somebody pulling out of a social interaction to check something on Google.
There are some things where a "better interface" would allow technology to slide into the background. A simple example is a good GPS unit with great voice recognition rather than having to type in a start and end destination. Maybe with a see through display on your car's HUD rather than on a small screen set somewhere out of the "normal" view of the driver so that all you have to do to see your next turn is change your focus point, rather than look away. Taken even further your car drives itself and you no longer have to worry about the road - you can interact, safely, with the passengers in the car, or over a cell phone.
That said, we're single taskers. If I'm composing a note to billy, be it with pen and paper, on a cell phone's screen, a keyboard, or a device capable of reading my thoughts so I don't have to talk, type, or write, I'll still be concentrating on that note. I may *try* to talk with people around me, but then both my conversation and my written note would obviously suffer.
There are a couple ideas that excite me, but neither really solves the problem of ignoring those around us in favor of our toys. One is the idea of an augmented mind via a neural interface. This is a long way off, but having the "hud" behind one's eyes, rather than something strapped on top of them is kind of exciting. Add in a computer as part of one's brain, and some wireless technology built into that, and I could see a realm where you get the best of the binary and analog worlds when it comes to thinking. The wireless bit would allow one to offload tasks that are too big for your built in computer, but just having the built in computer offers some cool vistas. There are just some calculations that computers are better at than humans, and being able to think in both modes would be quite exciting, I'd think. I don't know if we'd ever really get to the point where the silicon, for advanced interactions like programming or mathematics ever actually "feels" like a part of the brain, or if it'd just be a much faster interface than keyboard and monitor. I could see having sensors, like a compass, that the brain simply uses, similar to the interaction with our eyes or ears.
Another idea I like to kick around is a companion device along the lines of the movie "Her", with an optional robotic avatar ala Persocoms in "Chobits". I'm not talking the end game in either show where the computer surpasses us, but an actual companion device. It could be as small as a Furby, or as large as a human. (Technically it could be any sized, but I don't think we'd want anything bigger than that unless it's a vehicle we're riding in or a domicile.) Secretary, personal trainer, a more "natural" interaction with you and the people around you than typing in a screen, etc. You'd run into strange situations where an adult might become emotionally attached to the device. Asking something with even a hint of a personality about something that's available online could be a lot more natural than breaking off a conversation to look it up on one's phone, and a computer *would* be able to multi-task like that without breaking the illusion that the conversation is the center of its world. Japan's already looking into robots for elderly care, and if the robot had a bit of personality, it'd surely help there.
I guess if you combined the two ideas you could "ask" the computer side of you to Google for the currently tallest building in the world, or how close we are to getting long chains of carbon nano-tubes for a space elevator with a minimal amount of interruption and have it return the results to you. . . but even then, the regular gray matter would need to take some time out to send and receive the data from the computer half.
I think an external thing that's everyone could talk to, as a part of the conversation, rather than something one person does away from the conversation, could be more natural. Cell phones aren't that far from it. They have voice recognition and offline processing. Offload the computing and get a bit of rudimentary "personality" into it, put it into speaker phone mode, and it could join the conversation rather than taking people out of it.