Ronda Hauben on 20 Sep 2000 18:56:55 -0000


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: <nettime> The Age of Spiritual Machines (Review)


Alan Sondheim <sondheim@panix.com> writes:

>I may be way out of place in this debate, not having read Kurzweil, but to
>assume that a computer can only do algorithmic activity seems problematic
>to me, founded on older models of computation. For one thing, break the
>algorithm and feed the results to parallel neural nets - the easiest way
>to break is to use random numbers (not pseudo-random), for example based
>on radioactive decay - it doesn't matter.

When I heard Kurzweil speak ( I haven't read it his book.) he was
basically talking about algorithmic activity.

Certainly there has been an effort in computer science to make it possible
for the computer to do heuristic activity, for example I think that Allen
Newell's research has been toward that end.

And I have the impression that this is still a hard area.

But the effort Newell was making was not at the high end of the spectrum
of heuristic activity that LIcklider saw as the human ability that the
computer couldn't match.

Licklider was referring to the end of the spectrum of heuristic activity
where the question hasn't yet been clarified. He wanted to have the
computer be able to help in this, but also the human was then necessary
(and as far as I know still is) in this edge of the spectrum.

Newell, at the time (which which a while ago) was taking up the piece of
the spectrum of when the question had been clarified, then what could the
computer to do be helpful in the process.

Thus Licklider and Newell were collaborating in their research as
Licklider was fostering the human and the computer to be able to
collaborate.

Kurzweil's point is that in 20 or 30 years the human will be obsolete and
the computer will do all the human can do as far as intellectual activity.

Licklider was saying that the human and the computer should collaborate.

It would be good to see computers being able to formulate the questions in
hard problems.

But also Norbert Wiener points out that once a machine can do human
intellectual functions it is crucial to realize that another part of the
problem is to be able to formulate human goals. That in the human this is
a process where there is continual rethinking of the problem and requires
intuitive activity.

Wiener proposes studying "mixed" systems - ie. human-computer systems as
there you have the benefit of the human and the human brain with its
millions of years of development and of the computer and its amazing
capabilities.

>I personally see no limits whatsoever on machines, beyond the obvious of
>packing density, noise, tunneling, etc. in the hardware; the theoretical
>limits are enormous. And as the machine becomes increasingly responsive to
>external signals (temperature, human inputs, hearing/vision, etc.) new
>elements will be added - almost an inverse of Merlin Donald's thesis, in
>which the organic is extended into the machinic - in other words, the
>machinic will extend into and throughout the environment. To speak of
>algorithmic behavior in this context is to look back at Minsky's percep-
>tons as a guide to what we now know about neural functioning.

Licklider welcomed all advances in machine capability.

Licklider funded basic research to make it possible for the computer to do
heuristic activity.

The advances of our time, in good, part come from his program to advance
the human and to advance the computer and to faciliate their
collaboration. And to foster computer faciliated human to human
collaboration.

Do you disagree?

And I propose that in trying to understand what is a fruitful computer
science research program for the next 30 or 40 years that it is important
to sum up where our advances have come from, and try to determine the
implications of that toward where funding and personnel should be put in
the next 30 or 40 years.

I don't see that happening with the funding sources in the US
unfortunately.

And meanwhile Kurzweil is given widespread publicity and invited to speak
at Expo's like the PC Expo in NY about how computers will replace human
brains in the next 10 to 30 years.

And there is no support in the US for a review of Licklider's research
program and how that has been so fruitful over the past 40 years.

In fact in the May 2000 issue of Communications of the ACM, there is an
article by Davide Tennenhouse, who is now at Intel. But he was an office
Director at DARPA and in the article says that he wrote the article on the
basis of conversations he had with people at DARPA. The article is titled
"Proactive Computing"  And in it he says we should declare Licklider's
program of human centered computing a success and end that research
paradigm and adopt a new one. And he doesn't give any indication of
understanding or being interested in what Licklider's program was nor what
made it so fruitful.

However he is urging the computer research community to abandon it.


Ronda
ronda@panix.com
http://www.ais.org/~ronda/
http://www.columbia.edu/~hauben/netbook




#  distributed via <nettime>: no commercial use without permission
#  <nettime> is a moderated mailing list for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: majordomo@bbs.thing.net and "info nettime-l" in the msg body
#  archive: http://www.nettime.org contact: nettime@bbs.thing.net