mori

If it's like smoking, but safe, then who cares if less quit?

The projected amount of damage done by means of people hurt by not quitting ecigs needs to be higher than the projected amount of damage done by means of people not switching to ecigs for this to be alarming.

What I want to know is whether there's collaboration with MIRI. On safety, especially.

IMO: teaching people to appreciate culture or be good citizens is something that is nearly impossible (at least with the tools we have now). As it stands, it'd be a waste of everyone's time and money.

There is a lot of talk about how good it is to do this- yes, of course. But there is little talk of whether it is possible or how to make it possible. Comprehension of, say, math: that's something that can more-or-less be measured by a test. How do you measure someone having the attitude of a free thinker? Until practicalities are talked about, I can't see any of this as anything but posturing. There should be pushes for investigation like CFAR does.

Compassion is the feeling compelling on to help others, no? In that case, the best thing to do is to help others as much as one can- that is, as effectively as one can.

His comparing of logos and the "free expression of a skilled painter" is honestly ridiculous. They are optimized for completely different things!

What, preference utilitarianism? Afaik, that does get discussed, especially in the Fun Theory sequence iirc. If you want some more discussion on torture vs. dust specks, I think nostalgebraist on tumblr had a discussion of it with slatestarscratchpad (again, iirc).

Well, yes? If Roko's basilisk was true, it'd be unethical to spread. It turned out to not be true, but if Roko was reasoning from the premise of it being true, or even possibly-true, then the action was still unethical.

If it's smart enough to be a threat, I wouldn't be surprised if it's smart enough to avoid getting EMPed.

If you're looking to support AI safety, the Machine Intelligence Research Institute is researching it.