Re: Singularity and AIs -software

From: John Casey (
Date: Thu Jan 31 2002 - 15:20:44 MST

>> Gregory Bateson used to say, I think, that intelligence basically was
>> ability to find the pattern that connects two objects or events. I'm
>> why we've had so little luck devising programs that can do this.
>William Calvin would probably argue that it derives from the
>interaction of two thought patterns overlapping one another
>in the brain.

And I'd argue that for the purposes of this discussion the two statements
are functionally the same. Both are metaphors. Subjectively there's no
difference between anobject or event and a thought pattern about that object
or event. But either way I think we understand each other.

>Be careful -- there is probably a selection effect in that "Westerners"
>may have been higher levels of novelty-seeking genes. We discussed
>this on the list a couple of months ago.

Sorry, I only just rejoined the list (had to quit because of the high volume
of email!). But assuming it's proven that such genes actually exist (I'd be
ready instantly to agree the presence of cultural memes deriving from
innovations in the keeping-my-ass-warm field), does it make much difference
why, if the strategy consistently wins in the paradigm of social games we're
currently playing? Novelty and innovation aren't that different, after all.

BTW are there archives anywhere? I'd be interested in reading up on why
novelty seeking would be prevelant in cultures of European descent.

>The problem is -- do they "connect" from a common sense perspective
>of being innovative or useful or are they simply "interesting"?
>The connection of "sad--cow" generates an entirely different
>pattern in my brain than the two patterns generated by "mad--cow"
>which is different from the somewhat fuzzy "cow--mad".

Not every innovation is practical, nor does it need to be in a market-driven
economy. Demand will simply weed out the crackpot inventors and leave the
really useful innovations. I'd go further and suggest that without some
inherent psychic encouragement that it's okay to come up with impractical
innovations, most of the practical ones wouldn't have come into existence.
It's statistics - make a hundred new ideas, one will be really useful, so
much so that the idea's value will exceed the expenditure of energy required
to come up with it.

As an aside, in Singapore there's a word, "kiasu," which signifies being
afraid to lose what you already have. Singaporeans joke that this is a
national characteristics, and it manifests as reticence, risk-aversion,
unwillingness to stick your neck out. The same pattern exists all over Asia
in varying degrees. But an unwillingness to appear foolish occasionally will
clearly obstruct a will to innovate.

On the other hand, put a computer that can fail again and again, and at the
end of the process produce an innovative idea, into the hands of people who
don't intrinsically have the emotional or cultural wherewithal to do this
themselves, and innovation loses its excess value. Singapore becomes as good
as the US at designing software, making movies, inventing risky financial
instruments, and in order to maintain the Empire, the US has to move up a
level (to what?).

>As someone's sig says, "everything is easy once you know how to do it".

Right. And surely the notion that dumb luck in combination with an ability
to interpret unexpected results can be brought to bear. That's the point - a
computer that can generate possible connecting patterns (or, okay,
overlapping ideas) and test them out for practicality to a particular task
is a computer that, for all intents and purposes, can innovate. And I'd
suggest that the day we can produce these cheaply enough that Singapore (or,
perhaps Tonga) can buy one is the day of the Singularity.

>It is questionable whether humans could ever properly wire together
>100 million transistors -- but there are computers that can accomplish
>that clever trick. What one needs are the metrics and heuristics
>that generate useful patterns and efficient means for recognizing
>good combinations in specific knowledge domains. We are slowly
>(very slowly in most cases) managing to accomplish that.

Well, what do you think? Will it always come in slow, small steps? Or might
it be possible to use the process self-reflexively so we can accelerate it?
I mean, we don't have all day.

This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 13:37:37 MST