second hand translation (was: [alicebot-aiethics] Interesting Article)

[Noel Bush]

Hmm, thanks for the translation.  I am not sure if it eliminates the
confusion about Hawking's point of view -- in any case it makes it even
more disturbing.  It would be interesting to know the reactions to these
statements in Germany, where public discourse about such issues as
genetic engineering of human beings tends to be somewhat more elevated
than in the US (where people seem to have almost no knowledge of
history).

Hawking is a big advocate of the anthropic principle in physics as a
necessary "third pillar" to make cosmology a "real science" (his words
-- see http://www.hawking.org.uk/lectures/quantum.html).  From my very
naive understanding of this principle (and of quantum mechanics or
cosmology in general), I guess I can see how the beliefs he's stating in
this article, although they are way underarticulated (probably the fault
of the interview scenario, not Hawking), might follow from the anthropic
principle.

If you accept that part of the necessary explanation for why there are
cosmological constants is that we are present and conscious to perceive
them -- something that sounds backwards or tautologous but, I can only
take on faith (which certainly isn't science), is somehow rigorously
borne out by such arguments as in the article above -- then maybe you
can at least get the sentiment that is being expressed by
Hawking...which is something like "we have to find ways of being better
people", in which "better" implies virtuousness, some standard of
morality, whatever.  I think it is very reasonable to say that much of
the worldview that we have encoded into the machines we've built is
absolutely corrosive -- just soulless, even "evil" in some way...and I
am not a "technophobe", but I do believe that technologies are
transformative to humanity, and that it's possible to build a technology
that is transformative in a very damaging way.

This humorous (or offensive, depending on your take) article --
http://www.theonion.com/onion3123/hawkingexo.html -- parodies the idea
of a robot body, and ironically (for us on this mailing list) pins the
parody on Hawking.  Now, whether you laugh or cringe (or both) at this
type of humor, you have to admit that this keys in on a kind of
cybernetic dream that many people have...and on the assumption that
human bodies are weak and useless, machines are meant to enable us to
have unstoppable powers, etc.

But what about unstoppable powers?  Not the unstoppable powers to "rip
open enemy tanks like they were nutshells", as in the Onion piece, but
the unstoppable power to, say, do very fast mathematical computations.
Is it possible that certain magnificent capabilities like these can
actually corrode our humanity?

Many of us have probably had a teacher bemoan the present-day inability
of most people to do much more than extremely elementary arithmetic
without the aid of a computer.  One side of this argument is: you ought
to understand how to do arithmetic, even if it isn't practically
necessary for you to perform it yourself.  Another side is: why bother,
if the tools are already there?

There is a deep debate (I don't mean "deep" like "profound", but "deep"
like "mostly buried") in the AIML community about what kinds of
capabilities ought to be included in AIML.  AIML is, from the original
point of view, meant to model one idea of how the human brain works.
For some people, this model looks absurd because it doesn't incorporate
replicas of structures that we supposedly "know" are essential to the
functioning of the brain: there is no neural network in Alicebot/AIML,
for instance.  But for other people (I am in this camp), the model is
profound precisely for this reason, because it actually assumes a
possibility that we can model our own cognition in an accessible way,
rather than by resorting to building black boxes.  So, if you are
designing AIML (which is ongoing, in some sense), do you include an
ability to perform any arithmetical computation?

I would say no, for reasons posted here:
http://list.alicebot.org/pipermail/alicebot-archcomm/2001-September/0007
54.html.  Now, how does this bear on the problem of "unstoppable
powers"?

Well, if you take it as a sine qua non that intelligence includes the
ability to perform, internally, complex mathematics, then you are
ultimately setting a bar for humanity that will *require* that we all
have this ability, modeled in the way that computers do mathematics.
But what would it do to our consciousness if we did mathematics in the
way computers do?  It would mean that way back when, some engineer at
Intel or wherever decided how you ought to *think*.  If a fast
mathematical processor becomes so wedded to your brain that all
computations automatically get routed to it, then actually you cease to
participate in the intellectual discourse of humanity.  The human
interest in mathematics is not simply the interest in efficient
computation.  The study of mathematics leads to some very essential
philosophical questions.  The study of mathematics is in fact the study
of some aspect of what is taken for granted, and how so (and what else
might be).  Remove that, replace it with a floating point processor, and
you've put a big roadblock in the way of real thought.

In general this is a pattern we can see: cybernetics misunderstood as
mere technological extension of "capabilities", but missing the point of
"the human use of human beings" -- that is, the feedback part, the part
that is about giving us further insight into ourselves.

It would be really interesting to understand from a deep cosmological
point of view how failing to pursue this deep understanding of ourselves
might (in some theoretical space at least) suggest a kind of evolution
of the universe into a very undesirable place to "be", or even a place
in which "be" is stricken from the vocabulary of sensible concepts.
Surprisingly, a minimalist view like AIML simultaneously suggests that
"to be" may be an illusion, but that the illusion may be the most
meaningful thing of all.

Christian wrote:
> alright, I have a copy of the original article.
> I hope I will not make significant mistakes in the 
> translation. I also noticed, that the word "Ubermensch" is 
> actually mentioned...
> 
> The interview was done by Michael Odenwald for the German 
> magazine "Focus" and appeared in Issue 36/2001. The title is 
> "Ich konnte mit Einstein und Newton pokern" ("I could play 
> poker with Einstein and Newton", an allusion to an episode of 
> "Star Trek - The next generation" starring Stephen Hawking as 
> himself, playing poker with Einstein and Newton)
> 
> I will only translate the relevant part:
> 
> Hawking: "[...] Today aggression has has had its day as a 
> vital instinct. But darwinistic selection works way too slow 
> to eliminate it from our genes. I see our only hope in 
> genetic engineering."
> 
> FOCUS: "So, are you pleading for the creation of an 
> Ubermensch to take the place of the homo sapiens?"
> 
> Hawking: "By specifically editing the DNA we could increase 
> its complexity and thereby improve man. It will be a long 
> process, because one has to wait 18 years per generation to 
> notice the effect of genetic engineering. Nevertheless we 
> should choose this way if we want biological systems to 
> remain superiour to electronical systems. Unlike our 
> intellect, computers double their capacity in every 18 
> months. Thus the danger of [computers] developing 
> intelligence and taking over the world is real. So we must 
> develop technologies that facilitate direct connections 
> between the brain and computers as quickly as possible, so 
> artificial brains contribute to human intelligence instead of 
> turning against us."
> 
> 
> I hope this clears things up a bit...
> 
> Christian
> 
> _______________________________________________
> alicebot-aiethics mailing list alicebot-aiethics@list.alicebot.org
> http://list.alicebot.org/mailman/listinfo.cgi/alicebot-aiethics
>