From: Leonardo Wild (dlwild@access.net.ec)
Date: Mon May 19 2003 - 23:30:12 MDT
Hello,
Perhaps there was a mix-up of sorts regarding my reference to SIAI
looking for money because, in order to do what SIAI seems to be wanting
to do, it doesn't take a deep analysis (I agree) to see that without
money that's more or less impossible (under present circumstances).
> Cliff Stabbert wrote:
> Hi, you wrote:
>(...) We may be intelligent, but our intelligence doesn't
> LW> seem to be willing to create an awareness and hopefully understanding of
> LW> one of the most widely used man-made contraptions (money).
Yet the discussion was about a flawed analysis in regards to
friendliness, and the connection to values, and more or less such
discussions do not really mention money as one of the core elements in
society by which what we "value" gets measured, creating a set of values
that are intimately yet subconciously linked with that tool by which we
"measure," "store," and "transfer" that which we consider valuable.
> Yes, as Douglas Adams points out, it's just little colored bits of
> paper, it's all just symbols, etc. (...)
I take this is more or less in jest. Otherwise, your bringing up Douglas
Adams as your reference to speak about "money" would be a completely
unscientific reply to my longwinded paragraph. Up to sl4 standards? I
mean, that's the reason I had to step back for a month, for bringing up
issues refering to authors that are not considered, by certain list
standards, to be real scientists.
I'm not defending authors nor scientists, and I like Douglas Adams'
work, and his references to money are quite on the mark in many ways,
yet I would never have dared to bring up Douglas Adams as my "scientific
proof" for anything related to the issue of money. But I can understand
that being new to the list my posts are more closely looked at and any
half-degree deviation from the purity of science will be scolded, and
that those who are better known and have had a chance to prove
themselves in their knowledge-ableness have more leeway. So be it. But
it just points back to the issue that I seem to have understood Bill
Hibbard as trying to make, i.e., that we need to further define what we
mean by friendliness (and I add, "Not just that") in order to be able to
agree on something, especially on things as volatile or bound to
interpretation as values. As an example, you wrote:
> So your longwinded paragraph above might be of interest to those
> inclined to sit back and wonder about the meaning of it all, but
> SIAI's mission is to achieve things, not contemplate.
I see that you don't value longwinded paragraphs, or that your value of
them is lower (if not in negative) than short paragraphs that go
directly to the point. Also, the sentence says that "wondering about the
meaning of it all" is, more or less, a waste of time to SIAI's mission.
It directly implies, too, that "contemplation" is not a worthwhile
activity. So, for SIAI's mission, it is put into the box of the "bad
stuff" that will not allow SIAI to reach its goal.
Let me be clear here. I'm all for the need for friendliness in an AI or
else, as Eliezer clearly points out, we'll be considered disposable.
Unnecessary? Not "good enough" to allow lurking about? It's about goals,
right?
If you have a goal, and you want to carry it out, it would be an
"intelligent" thing to do to analyze what it is that you are after, as
well as what means you need to make that possible. So it was said: "We
need money." Okay, that's clear. But it is not intelligent, really, to
go after the money without knowing what it is. It is philosophy, yes, to
go after something you don't know what it is and, when you realize that,
wonder why you are doing it? The funny thing is that most people in
today's world arrange their whole lives around the "acquisition" of this
thing we call money. But where you to ask them: "What is money?" they
can't give you a coherent answer. They tell you how you can get it (by
working hard, by gambling in Las Vegas, by robbing a bank, by investing
in the stock market, etc.), but all those replies will _not_ tell you
what money is, they will tell you how you can acquire them and, perhaps,
a whole study can be made in terms of which kind of activity is the most
suited for reaching your goal of having money.
And Gary Miller replied to that very same posting with:
> GM: If on Tuesday they have a paranoid psychotic
> AI they restore him to Friday and try to determine what made him bonkers! Maybe
> they Should have left him read Gandhi's autobiography instead of of Richard
> Nixon's. Or maybe they just forgot to assert Gandi good, Nixon bad!
>
> I did not of course try to define all the inputs that are required to
> make a normal, social, and mentally healthy FAI or even attempt to define
> what one of those is. I leave that to the behavioral psychologists and knowledge
> engineers.
It wasn't a matter of defining all the inputs, but rather of defining
what it is that we mean by "friendliness." Replies are Gödelian in
nature, just like waiting for an AI to say that it is intelligent. In
fact, Gödelian logic works the other way round. In order for an
intelligence to be intelligent, it must prove that it is _not_
intelligent. That's all very fine with intelligence, because its
function is completely different than, say, friendliness. Friendliness
is not a function, it is the result of a need, so it would completely
undermine the goal of creaing a FAI to wait for it to prove that it's
not friendly.
It so happens that the _functions_ of "money-as-we-know-it" are
contradictory in nature, to the point where, due to this contradiction,
the behavior it incites in people who, as Cliff Stabbert said: "(...)
these symbols are real to the extent that other people behave as if they
are real -- and in this universe, on this planet, at this time, most
other people do behave that way." But _what_ behavior is that? Not just
that money is real, but much more than that; one of the elements is an
inherent induction of a kind of behavior that we might call "unfriendly
towards life in general." The kind of behavior money induces in people
is akin to the pheronomal (chemical) communication in insects, where
ants, for example, will be bound in their behavior due to the
semiochemicals secreted by the other individuals. So your assumption
that money is what it is because we look upon it as real and that
therefore, since it's real, we shouldn't bother wondering what money is,
this kind of attitude being exactly one of the fallacies in the
friendliness analysis.
If behavioral psychologits and knowledge engineers are left with the
definition of friendliness, they will do whatever they feel is right as
long as _they_ get paid for it. And who will pay them? Those entities
that have the money. And what will they pay for? To further their own
goals. And what goals will that be and how can you ensure that _their_
kind of friendliness (and the target of that friendliness) will make the
FAI "safe" for humanity at large ... which is the goal of Eliezer, as
far as I've managed to gather.
Let me repeat what Cliff Stabbert wrote:
> CS: SIAI's mission is to achieve things, not contemplate.
I don't know if you realize what you said, there. That SIAI isn't into
contemplation. And written in a way that:
a) it assumes that I am not a "doer" and merely a contemplator who
doesn't achieve things and
b) that contemplation isn't really something worth the time of those who
are part of SIAI.
You are probably referring to certain _kind_ of contemplation, though
not at one of the first definitions of it: 1."The act of looking at
something intently." Nor of yet another important definition of it: 4.
"the act of looking forward to; intention or expectation."
It is part of human intelligence the ability to contemplate ahead and
not do as certain insects do: plow ahead until they reach an obstactly
and then try to feel their way around/over/under it. Upon reading once
again my "longwinded paragraph" I found that you probably didn't catch
onto many of the subtle comments; since money is not of interest to you
except for _how_ to acquire it in order to fulfill your goals, the wider
picture seemed too undefined to understand. But let me say something in
regards to money: When and if a Singularity comes into being and
develops into a FAI, one of the _first_ technologies it will take a
closer look at is man-made contraptions that are at the core of our
"unfriendly" behavior towards each other and the environment that
supports us. Using Bill Hibbard's very clear statement: " (...) choosing
between different value systems requires base values for comparing the
consequences of different value systems."
The value system we base our _other_ values on (such as friendliness) is
the system by which we acquire what we need. That is, our so-called
economic structure. Which is, basically (and in the present time), a
structure based on a tool we call money.
Friendliness is a behavioral pattern that has evolved because, through
it, it ensures the survival not only of the individual, but of the
species. Yes, Philip:
> Frans de Waal ???
> http://www.emory.edu/PSYCH/Faculty/dewaal.html
... is the author I was thinking of. We can see that certain behavioral
patterns that were considered arising in the human only are, in fact,
part of the strategy of survival of many species.
The other interesting aspect of money is that it is being created
through a collusion that has left out the original consensus of "what
money is." In other words, the "social agreement" is one of imposition.
I'm not trying to convince anyone to look and contemplate into the
issues regarding money, but I can assure you that there are thousands of
places in the world where there are in fact people who aren't just
contemplating the issue of money, but also doing something about it.
Not that SIAI must get into that. I had just brought it up to show how,
something as widespread as money, which is at the core of our base value
systems (and at the core of the problem Eliezer and others are facing
when wondering what kind of company or organization will have the
capacity to create an AI), is something that we all take so much for
granted that we don't even realize how our lives are ruled by it nor why
that is happening the way it is.
If Friendly AI is mean to deal with the issue of Poverty, then it _must_
look into money, since poverty is, per definition of how economy works
today, _caused_ by how money works (not just what it is).
But enough for today.
Best,
Leonardo Wild
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT