I am surprised that no one brought up the following point yet (this being LW after all): if you believe that strong AI will be developed before applications for the theoretical research are found, then, from purely practical point of view, you shouldn't dedicate your efforts to the research. It is more efficient to assist the development of strong AI instead of doing the research yourself.
If you have python you can run the following script which will subscribe your account to the aforementioned users: https://github.com/afiodorov/mass_twitter_follow