Re: An essay I just wrote on the Singularity.

From: Perry E. Metzger (
Date: Sun Jan 04 2004 - 20:36:38 MST

Samantha Atkins <> writes:
> On Fri, 02 Jan 2004 17:06:29 -0500
> "Perry E. Metzger" <> wrote:
>> Lawrence Foard <> writes:
>> > On Fri, 2 Jan 2004, Perry E. Metzger wrote:
>> >
>> > [...]
>> >
>> >> So what sort of strategies does evolution favor? Quite a number of
>> >> them, actually, but none of them can be characterized as "pacifist".
>> >
>> > Not pacifist, but also not unreasoningly or overly aggressive.
>> If you think ours is not an extraordinarily vicious and aggressive
>> race, I direct you to the nearest factory farm or slaughterhouse. Most
>> people are fully aware of where their food comes from, and yet we (in
>> general) have very little compunction about continuing to eat it.
> How the race on average is and what one's own choices are now and on
> the other side of a Singularity are very different things. For a
> time I was vegan. It seemed moral and reasonable on multiple
> levels. Then I decided to add back in at least some animal protein
> (mostly seafood) as I seemed a bit sharper and more focused.
> Eventually I got to today where I limit intake of most meats but am
> not as adverse to them. Some of the premises of my earlier
> valuation changed. In other places, like most humans, it was to be
> honest simply more convenient to not buck the tide. Does that mean
> I am vicious and agressive? No, I don't think so.

Certainly many animals would very actively resist if you attempted to
kill them personally to make them into a meal for yourself -- you'd
have to use fairly brutal means to carry it out. As it happens, you
get proxies to do it for you, but you might want to ask if, say, you'd
feel quite the same way about how peaceable you are if you had to
kill the steer yourself with a machete.

A PETA member might claim that you're vicious and aggressive for
feeding yourself this way when alternatives exist. Myself, I make no
moral claims in this instance, as I don't make unqualified moral claims.

>> By the way, I will repeat a contention I've made pretty frequently,
>> with a new example. There are people out there that say "meat is
>> murder". There are people who enjoy a good steak. Which of them is
>> "morally correct"? What is the experiment we can conduct that will
>> answer this question? I contend there is none, and that there is no
>> answer because absolute morality is an illusion.
> The form of the quesiton is probably too imprecise to be answerable.

I think it has a fairly precise consequence. If meat is murder,
Mr. F. A. Intelligence will be called upon to prevent it. If it is
okay, presumably he'll go about assisting his charges in improving
their abattoirs. So, what is the experiment we conduct to determine
which path Mr. F. A. Intelligence should follow?

By the way, I'm not actually that concerned with the animal rights
question. As a couple of people have accurately noted, it is merely a
proxy for a giant range of moral quandaries our hypothetical
Jovial Neighborhood Superintelligence might be faced with.

> > Our race intelligently tends to stop being vicious only when self
> > defense is likely (as in when dealing with other humans) and tends to
> > be pretty brutal when dealing with life forms unable to defend
> > themselves effectively. Baboons can be pretty nasty creatures, but as
> > we've got the cages and they don't, guess who is used in who's vaccine
> > experiments?
> We are a bit removed from raw primate
> behavior and our acheivement of transhumanity requires us to become
> even more removed from those roots. Otherwise I would have no
> interest in the entire enterprise. Super-powerful baboons are not
> my idea of a good time or soemthing I would want to have a hand in
> brining into being.

I didn't mention super-baboons, although we do in many ways resemble
that. My point is that we have no absolute moral calculus with which
to ascertain, precisely, the answer to questions like "is it okay to
experiment on animals" or "is it okay to execute a vicious criminal"
or even "is it okay to cheat on your wife". Indeed, I'm not sure that
we have a way of unambiguously saying that meat is murder when the
meat in question is human.

That makes the Friendly AI a bit difficult -- we don't have
particularly good ways of determining, a priori, what Friendly might
even *mean*, and certainly not without enough precision to know what
we'd be in for.

>> That's not to say, by the way, that I draw any sort of absolute
>> morality under which gluing the feet of geese to the bottoms of cages
>> and force feeding them to make their livers particularly tasty is
>> "evil". I don't believe there is an absolute morality, so I'm not
>> going to pretend that it is somehow "wrong" to turn geese into foie
>> gras. However, you would be pretty much incorrect if you thought the
>> geese weren't pissed off about the process -- it is very obvious that
>> they're miserable. (One might argue if animals can suffer, but the
>> arguments used to claim they can't could just as easily be applied to
>> other people -- I'm going to stipulate that if a cat screams in pain
>> when you stick it with a hot iron that it is in fact suffering.)
> There is no argument about the obvious fact that most animals can
> suffer. Having heard how foie gras is achieved I most certainly
> will not eat it. By my standards it is wrong.

Ah, but by my friend C.'s lights, it is a tasty meal and there is
nothing wrong with it at all. By what experiment in Absolute Morality
might I convince him of this? Or are you willing to concede that your
standards and my friend C.'s standards are not comparable by outside
means -- that is, that each of you has a purely relative morality
without claim to absolute moral truth?

>> > You will often see that predators have evolved strategies to
>> > avoid actual aggression, instead resolving disputes with a proxy
>> > aggression to assert dominance and territory.
>> *With each other*, since two members of the same species have nearly
>> the same capacity for violence. It is rare that you will see
>> cheetahs negotiating with herds of antelope to try to non-violently
>> settle their "disputes".
> Fortunately we have a much greater intelligence and ability to find
> other ways of being with one another than cheetahs do.

Fortunately in that we'd end up leaving fewer offspring if we did not
find such such means to deal with EACH OTHER. However, it appears that
no such consequences arise from, say, clubbing a seal and then turning
it into a coat, or taking a turkey and raising it in a miniature pen
and then cutting its head off, so many of us do such things.

We pretend to ourselves that there is something "wrong" about taking
our neighbor's child and eating him, but that it is "okay" to take a
calf and eat it. We further pretend that "wrong" and "okay" are
written in the fabric of the universe. The truth, though, is that we
simply are assigning "good" and "bad" values entirely relative to our
own interests, and that the universe itself is entirely value
free. There are no mile high letters on the far side of the moon that
tell us the "Truth" with a capital "T". We just have a set of evolved
behaviors that we are comfortable with.

When we accept that what we like and dislike is purely a consequence
of evolution and resulting taste, we rapidly see the foolishness of
asking what is the "right" answer to the animal rights question or the
abortion question or any other such question. There is no "right"
answer, and the question is meaningless, and we can abandon trying to
settle such questions via the application of pure moralities and
instead ask things like "well, what are the consequences to us if we
sanction behavior X" instead of pretending there's a moral answer
sitting out there.

The whole system breaks down especially badly when we start dealing
with transhumanity and posthumanity. Sadly, some people are pretending
to themselves that, deep down, if there is no God who created them in
his image, they can at least create a new God in their own image. I
shudder to think of the consequences.

Perry E. Metzger

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:43 MDT