I really like the product you've made, and use it regularly. It might be helpful for your user base, and the project as a whole if one responded with a little more than a shrug. Facebook may be more willing to listen to you if you contact them then random users in getting this resolved.
To expand on what phil is saying: It is by itself logically consistent for someone to decide that there's enough fun left in their life and enough stuff that they need to do that they don't want to die now even if they don't necessarily mind dying in the indefinite future.
Uncomputable problems are not necessarily NP-hard. You can for example encode a version of the halting problem such that the instances are simply very long. For example, consider the language L defined by k is in L iff k=2^2^2^n for some n and nth Turing machine halts on the blank tape (some reasonable listing of Turing machines). In order to feed an NP-hard problem to an L-oracle, one needs to ridiculously expand the length of the problem far more than polynomial length so this certainly can't be done in polynomial time.
I'm not sure what could be done that would be dangerous in the name of "ushering in the singularity." The worst plausible scenario is that we put money into research avenues that don't pan out. But that happens all the time anyways. The real worst case scenario is that someone gets too afraid of AI and goes all Sarah Connor. But that's not a likely scenario especially because most of the people who take the Singularity idea seriously are people who are generally optimistic about the long-term trends.
I'm not sure what could be done that would be dangerous in the name of "ushering in the singularity."
How about firing up a potentially runaway AI without a stable goal system? Releasing a virus to "improve" people in the releasers' preferred direction? The likelihood of these seems minor (and for the first, might not even been possible), but the consequences are so grave that they bear some thought and worry anyway.
