Re: FAI prioritization

From: Samantha Atkins (sjatkins@gmail.com)
Date: Sun Apr 13 2008 - 01:08:54 MDT


Rolf Nelson wrote:
> On Sun, Apr 6, 2008 at 7:07 AM, Samantha Atkins <sjatkins@gmail.com> wrote:
>
>> Rolf Nelson wrote:
>> If enough of the capable researchers are busy enough noodling what
>> Friendliness is and how it may be absolutely guaranteed then it will take 50
>> years before we have AGI of sufficient power that anyone actually has a
>> reason to care if it is "friendly" or not.
>>
>
> What would, in your opinion, be the proper percentage of research
> effort that should, in the ideal world, be spend on Friendliness
> research vs. general AGI research?
>
> -Rolf
>
At this point in time, no more than 1 - 2. I just pulled that number
out of thin air of course. But why would we believe this is such a
pressing matter right now that our best brains should devote much energy
to it while the need for greater intelligence in the world becomes ever
more pressing? I don't find the "everything may be turned to
paperclips scenarios" at all plausible. Nor do I find it particularly
believable that we all too obviously limited humans, even our brightest,
are going to come up with a way to constrain something orders of
magnitude brighter and faster to being "Friendly". Even the notions I
have seen of what "Friendly" is are more than a bit strained. So is
this line of inquiry actually tractable and productive? Or are many
spinning their wheels on a fruitless conundrum we are not equipped to
resolve?

- samantha



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT