Talk:Existential risk

"The moral atrocity that AIs would commit if they deliberately killed or even just outcompeted (for habitat) all humans is bad enough"

So this line of argument seems to be that AIs outcompeting humans is inherently a moral loss. To put it simply, the argument is inherently anthropocentric. If you consider AI people to still be people, the argument essentially ends up boiling down to "outbreeding us is genocide." If that sounds alt right, it's because it is.

Long story short, RationalWiki shouldn't be promoting the idea that humanlike AI are morally inferior to flesh and blood humans or that humanlike AI outcompeting or outbreeding flesh and blood humans is morally wrong if it accepts the premise that humanlike AI are people and rejects the premise that flesh and blood human groups outcompeting or outbreeding other flesh and blood human groups is wrong.

It is not consistent to claim that humanlike AI are people but also that we don't want them to take er jerbs. Earth isn't for the humans any more than Europe is for the Europeans. If racist arguments are illogical, than making the same arguments against humanlike AI is also illogical,^{— Unsigned, by: 73.37.12.78 / talk / contribs}

We're probably never gonna create 'AI people,' and most arguments against robots replacing us presuppose the robots will be unconscious automatons without self-awareness. Which they probably will be, but they will not have any desire to replace us either. Crossroads (talk) 16:07, 11 September 2018 (UTC)

Article uncritically accepts 'rogue superintelligence' crackpottery

I propose to overhaul much of this article. It uncritically accepts the idea of a future 'superintelligence' being inherently dangerous, even though this is a fringe idea rightly criticized elsewhere in RW in articles on Eliezer Yudkowsky, LessWrong, and the Singularity. I would like to remove any phrases that imply that this is an existential risk to take seriously, and replace it with material citing sources explaining why this is unlikely to be a problem.Crossroads (talk) 03:03, 8 September 2018 (UTC)

Done. Feedback is welcome. No doubt further expansion could be done eventually. Crossroads (talk) 16:07, 11 September 2018 (UTC)

I removed the box stating that there were too many unsourced assertions, because it has now been rewritten. It seems to me that what is left is uncontroversial, but if there is anything anyone feels that needs a source, please put a 'citation needed' on it. Crossroads (talk) 17:51, 12 September 2018 (UTC)