I wish to develop machines which exhibit general intelligence. Although inspiring, most of the achievements in artificial intelligence have been very domain specific. It is not that we as a community never aimed for general intelligence, but it's been the case that we have not been successful. To me, it appears that understanding human brain is the holy-grail to this end.
What I think would drive the future:
I believe research in understanding the computational principles underlying the working of human brain would pave path for major technological breakthroughs. Brain is an intelligent system which is not only extremely energy efficient, but also very adaptive and allows for massive parallel computation. It was the von-neumann architecture which revolutionized our world in the past century - maybe it is time to move ahead now!
I personally like the work of Bruno Osshaulsen, Jack Gallant, Jitendra Malik, Geoff Hinton, Andrew Ng, Yann Le Cunn and Pieter Abbeel. Their research provides diverse and unique insights into understanding and building intelligent systems.