[extropy-chat] the question of self-termination

Eliezer Yudkowsky sentience at pobox.com
Fri Nov 19 21:58:45 UTC 2004


Damien Broderick wrote:
> My earlier post might be regarded as cruelly tactless at this sad 
> moment. But the fact and ethics of suicide need to be confronted. But 
> not now, not at this terrible moment? Well, if we scoff at the religious 
> for their death-bed and graveside evasions we dare not go into denial 
> mode ourselves over this aspect of our unyielding realities of life and 
> death. And while we hope to make death yield, still there will be those 
> who chose it (in whatever remediable or persuadable state of mind we 
> will never know).

You should know better than to invoke the unknowable.  Not all attempted 
suicides succeed.

> For a list as adamantly wedded to individual choice as 
> this one, suicide remains what Albert Camus called it more than 60 years 
> ago in the opening of Le Mythe de Sisyphe: "There is only one really 
> serious philosophical problem, and that is suicide."

I recall reading from a media story on suicide, that a study on suicide had 
found that very few failed suicides try again.  In the case of people who 
survived a jump, most reported regretting the decision almost the instant 
they started to fall.  "What the hell am I doing?" is the quote that sticks 
in my mind, and "As soon as I jumped, I realized that everything that had 
gone wrong in my life was fixable, except my decision to jump."  Is a 
decision that changes so predictably volitional?  If the report is correct, 
most people who jumped off bridges died involuntarily.  I can imagine a 
mind committing fully volitional suicide, a mind that has finished what it 
regards as its life.  I question whether this is the case for people who 
jump off bridges.  I question whether, in a more sensible world, the action 
would carry through without a check on volition.  What about people who 
commit suicide thinking there's an afterlife, or without realizing how much 
they might be missing?

How young is old enough to die?  6 years?  18 years?  100 years?

My personal opinion on this had previously changed - sometime before July 
2004 - to say that if I, myself, had to generate an opinion right now, I 
would require the same kind of driver's license to commit suicide as to 
engage in other radical self-alterations.  If I had to opine right now, I 
would opine that I'd grant anyone who could build a *Friendly* seed AI an 
unrestricted right to self-modification, not because it's enough knowledge 
to do it, but because it's enough knowledge to be scared, plus a 
fully-understood FAI system that you can use to check your consequences for 
you.  I don't know if I'd opine that there should be lesser tests for 
lesser abilities.  There's a strong argument for people's lives belonging 
to themselves as soon as possible, but on the other hand, "as soon as 
possible" could just as well mean, "pick a threshold such that not a single 
member of the human species actually commits suicide, unless they really 
mean it".  I had come, even before this point, to the conclusion that death 
is the worst thing in the world because it can never be repaired, never be 
made up for afterward.  Philosophical angst seems to me a lesser problem 
than death.

-- 
Eliezer S. Yudkowsky                          http://singinst.org/
Research Fellow, Singularity Institute for Artificial Intelligence



More information about the extropy-chat mailing list