I believe that just building an intelligence will not prove to be the most difficult task - it seems much more complicated to build it in a way that will not end up with humanity dead (or worse).
Friendly AI[1] is a bigger problem than General AI, and unfortunately it needs to be solved first.
[0] - http://lesswrong.com/lw/x7/cant_unbirth_a_child/
[1] - http://en.wikipedia.org/wiki/Friendly_artificial_intelligenc...
But I don't get the impression that it's something a lot of researchers are paying attention to. It's pretty much just the folks at MIRI (most famously Eliezer Yudkowsky, who posts on HN sometimes with id "Eliezer") and maybe FHI in Oxford. Or is there substantial work on this elsewhere?
That's what I meant (though maybe forgot to parenthesize it properly ;)); it's large because as I see it, it gets mind-bogglingly complicated quickly and branches into lots of different areas of maths, philosophy and computer science.
It is an active topic, though unfortunately I haven't seen much research in it aside from MIRI and Oxford guys. I hope this will begin to change, as topics of AI seems to enter the global consciousness again (thanks to Google and Facebook employing high-profile AI scientists, transhumanism becoming more popular in the media, etc).
(In a way that will survive self-modification of the AI's source code, of course. This is a very large and active topic of research.)