We also work on stuff that's going to help developers, like how to make compilers better, how to write a spec at the same time you're writing the code, how to only write the code once and reuse it many times and all sorts of things. We're developing new languages and new operating systems. Some may or may not see the light of day, but concepts from those get over into the products.

We have prepared a concept demo of what it would be like to have your own personal robot. Microsoft has a robotics studio and we're actually doing some consulting with some of the main robotics users, like car manufacturers and the like, and trying to make improvements there. In our labs we have a flying robot, it's had problems with its sensors and determining glass and transparency, we've had a few crashes with it! But we anticipate this kind of personal companion idea will be there by the time we get to the 2019 office vision, and I think that will be part of it as well.

I don't think the robots will have personalities, but they will be able to offer a lot of functions. To me, there's this vision that somehow we can create how human beings are, but at the moment we have a big enough challenge trying to get computers to be effective at speech recognition and turning them into something that can emulate human vision, which is very hard to do, and getting robots to speak in a way that's very similar. Right now, you're nodding your head, your eyes are flickering, there's all sort of cues that I'm taking -- how do we create this same experience within the computational. So, putting personalities in robots I think is nice for Star Trek and those other things, but we have so many other problems to get to, if we solve those we might be able to work on computer devices that have more human attributes, but until we can get computers to see, hear and speak better than they do today, and I'm talking about 100 or 200 times better, then maybe this will change.