I would like to address an aspect of this question that hasn't been mentioned yet.
Correct me if I'm wrong, but this is how I interpret this question:
I believe the NSA may have backdoored most computers in the world; how do I protect myself against it?
The only answer to this question is "You can't".
Why limit our threat model just to the NSA, though? Why not add Mossad, PLASSF, the Illuminati, SCP Foundation, ${InsertNefariousAgencyName}
to our list of threat actor? PLASSF seems especially plausible here, since most parts are manufactured in China.
I am saying this without irony - we the vast majority of people do not know what is actually going in the dark, clandestine world where politics meet intelligencies meet huge businesses. It goes without saying that there do exist unscrupulous, power hungry people and organizations that dwell there. We might not even know the names of all such agencies.
But whether they have backdoored most computers is irrelevant here. Even if you're right and the IME is a hardware backdoor that serves THEM - in the broadest sense of this world - and to protect yourself against THEM you disable the IME, THEM may have more, yet undiscovered hardware backdoors in your computer. And even if in some miraculous ways you disable these backdoors as well (or, more plausibly, just stop using any electronic devices whatsoever), then THEM will still have many more ways to exfiltrate your private data - including sending actual people to do what Harry Pfarrer thought was being done to him.
And if you do have enough knowledge / resources to actively oppose THEM, then you are Edward Snowden and are not asking such questions on this site.
While you cannot make THEM stop spying on you if THEM wish to do so, you might be successful in making THEM interested in your person by trying your best to hide something from THEM. Imagine you walk past your child, if you have any. Now you see your child immediately turn their back on you (in the literal sense), bend, cross their arms, clearly holding something and trying to not let you see what exactly - aren't you immediately concerned what the kid may be trying to hide?
You can defend against THEM by not sparking THEM's interest in you. The vast majority of people successfully do that.
Note: Again, I am not even trying to discuss any possible secret 3G communication in Intel devices. I am, however, questioning the validity of the OP's threat model. I realize that questions like these have important academic value and raise perhaps even more important privacy concerns, which is a very important (sorry...) subject of public debate, nonetheless I think that it still needs to be pointed out that the practical value of such questions (ie. "how can an average person such as me protect their privacy against this") is virtually nil.