I really can't make up my mind whether I think this is a good or bad question. No offense, because it is very interesting, indeed!
First of all. The question relies on some very controversial preconceptions. Namely that people treat computers as little humans.
Then it ask how we can achieve more of something we don't really know if we want (or consider it to be good UX). Do we really want to (team)work with computers (or any other device?) or do we want it to be a good transparent tool - just like any other convenient tool we use?
(Pictures from "A Brief Rant On The Future Of Interaction Design". A very nice read!)
Is the hammer the carpenters best friend? Or is it just another tool?
Is the iPad the students best friend? Or is this also just another tool?
Do we want the iPad to be friendly or do we just want this thing to have a high degree of usability? (30 years ago "user friendly" was the term for what we call "usability" or perhaps "user experience" today. The conclusion was clear: "user friendly" is misleading, because we don't need friendly computers, we need tools that makes it easy to accomplish the task we need to accomplish).
Ok, down to some answering :-)
To me, what you describe is just a consequences of good UX- and UI-design. A good interaction between the user and the "thing". If the user feels like he is in control, and that the "thing" does what he wants, then he will interact with the computer (or any other device) in a natural way.
The "Actor Network Theory" (Wikipedia article, YouTube movie) describes a network of actors (humans) and actants (eg. computers, mobile-phones or even hotel room keys). And I believe that Nass is mixing these two networking roles when he describes "computers as little humans". Yes, the actants have a place. And yes, the actants must be taken into account when designing and developing devices and software (especially devices/sw that affects the interpersonal interaction/communication). But the actants are not little humans.
(Illustration from www.clipartof.com)
How do we achieve a natural interaction then?
Well. This actually comes down to the very basics of UX knowledge. Best described (imho) by Don Norman in his famous "Seven stages of action" (Wikipedia article). A natural and good implementation of these gulfs (of execution and interaction) will lead to a natural interaction with the device. And in some circumstances this might lead to a subconscious comparison to interpersonal interaction and thus an simplified interpretation: "this is a human-like interaction, I must be dealing with a little human".
(Illustrations from www.dubberly.com)
Sorry if I'm a big party-pooper here. :-/
I haven't read Nass' book, so I really can't argue with his findings. They may very well be real and they are definitely interesting.
I do believe that we are aware that the computers are not human. I think we know that hitting the key harder will not make the computer understand better, or that hitting CE multiple time will clear the calculator better. We just do it, because this is how we are familiar with treating things.
So, bottom line. The computer/device has a job to to and it is an actant in the task we're doing. A tool. With a good exec/eval gulf, we will interact naturally - but still not expect them to be human. We still expect them to search faster and calculate better, but we don't expect them to understand feelings (like frustration) or interpret a picture.
The computer is just another tool, and we need create the appropriate tool for the given task.