Saturday, October 24, 2009

Could Terminator II come true? A true story

Background: a friend on a listserv suggested I should really collect my curious life stories in a book some time. I said I had thought about it, and decided against, in part because no one would believe it anyway...
He asked why, and I replied with an example:

==================================================
===========================================

Thanks much for the flattering comparison, Bob!

But no, I wasn't really thinking about Da Vinci. I was thinking about
hundreds of implausible stories.
So I should give you an amusing example. This story is one I **HAVE**
told to a number of people; it is more normal and mundane than most. But
even so, some would object, and I tend to doubt whether it belongs in
print. (I will not name names, but could. The naming of names is part of
the problem.)

I call it my "Terminator" story.

====================

Circa 1991, the International Joint Conference on Neural Networks
(IJCNN) was held in Seattle. Boeing was a major player,
so they arranged a couple of VIP tours of their aircraft plant with
lunch in the executive dining room.

Prior to that lunch, I was alert for any opportunity to really push
forward the neural network field.

I was delighted when I met a guy from McDonnell-Douglas who was
wrestling with the very tough problem of
how to assign missile interceptors to incoming missiles when there might
be hundreds of both. This was a problem that none of the usual baby
shortcuts could solve... so I spent a few minutes happily showing him
the flowcharts for how it
could really be addressed effectively, and planned to get back to him. I
was also delighted when I met a youngish Afro-American
researcher highly credible in neuroscience, tryuing to make connections
that would help him find real functional explanations of how things work
in the brain; I pointed him towards a new engineering research group I
planned to fund, and again intended to follow up.

Then at lunch -- one of the Boeing guys said: "There have been so many
of these new emerghing technologies like neural
networks down through the years. Someone really ought to do some kind of
study to help us predict which ones will really pan out and deliver and
which ones won't." As it happens -- back in those days I was Program
Director for two areas at NSF, neural networks and Emerging Technologies
Initiation (ETI). So I couldn't help repeating what I had heard from a
previous ETI program director: "Actually, we have done several studies
like that at NSF. One of them found a number of important predictors.
Some were things you would have expected -- like fields based on a new
breakthrough, or new crossdisciplinary connections. But some were a
little surprising -- for example, science fiction. Technologies subject
to good science fiction somehow mysteriously prospered more than you
would have expected. Bad science fiction predicts a strange withering.
By 'good' and 'bad,' I don't measn quality of writing; I mean whether it
is depicted as something good or bad."

Suddenly a loud groan came from one of the Boeing people. Me: "Why did
you just say... UUUUUGGGGHHHH?"
Him: "Talk about bad science fiction... have you seen Terminator II?"
Me: "Not really." (The title didn't sound like my kind of thing.)
Him: "Well, you have to. After what you just said, and what you are
doing, you really have to. It is your professional obligation."
Me: "Well, OK.".... (Then discussion of Star Trek and Data and other
subjects..)

Back in the office on Monday, I was very pleased to start the week by
signing off on an SBIR grant which basically started/created a new
company called "Neurodyne," which would finally create the first firm
base for a new stream of neural network technology, using a robot arm,
as the initial testbed application. And on Tuesday night I saw the movie.

As I saw the movie... it began as the human race was about to be
exterminated all because some yoyo told the missile defense people about
how to use intelligent decision-making neutral networks. Oops. The evil
robot about to destroy our last hope looked like a 50-50 morph between
me and the guy who really set up Neurodyne. Oops. Interesting to see
such a nice familiar face in a different role. And the key Afro-American
scientist working with Neurodyne in the movie was a dead ringer for the
guy I was just putting in touch with them. And it all started with a
ronot arm and a chip, and I just signed off on the robot arm the day before.

One person I told this to last week asked: "Didn't you wonder right then
whether someone was spying on you? It's so improbable?" My reply: "Maybe
for a millisecond.. but of course the movie had to be made long before
the actual events. Yet another major part of the movie was information
being sent back through time, and a woman who had access from the future...
and as soon as I put THAT part together, I really became creeped out."

I was well and truly shaken up for a few days, but gradually calmed
down. I told myself: "There is no way that computers could outsmart
humans to that extent, even using my new designs and methodologies,
without a really huge amount of sheer computing power. Von Neumann
estimated that brains operate at something like 10**15 or 10**18, and it
now looks more like something on the high end. We are 'way away from
that kind of horsepower know, especially when you think of the extra
horspepower needed to emulate variable connectivity." It really would
require a new chip as well as the robot arm project. So I calmed down...
to some degree.

Until about Thursday. On Thursday, I picked up the phone... and heard a
thick Germanic voice which really did sound just like Schwartzenegger.
"Hello, Paul!" (Picture the Schwartzenegger tone of voice.) "I am
calling from LA to tell you some good news.
We have this great new chip, and we are working with Neurodyne. We would
like to come and brief you on the new chip."

I probably need to start censoring at this point, even though it really
becomes much more entertaining and improbable after this point. It's
probably safe to say that I pushed hard for a new effort which funded
this extremely competent group to develop a new more near-term chip, for
use in solving air quality problems in cars, which was almost widely
deployed... until the Bush Administration got rid of the new air quality
regs which drove it. I later ran across a guy from the ballistic missile
defense organization, once by accident while riding the Metro and once
by accident in Circuit City, who happened to have a laptop handy with
powerpoint and video of how they were planning to use the more powerful
chip and showed it to me impromptu
right then and there. (But what I am leaving out...! And how many of
you have been innocently riding a Metro, a bit tired,
on a routine day to work, suddenly seeing a tall guy in a blue uniform
who could be right out of the movie, suddenly pop up a laptop and show
you real world terminator movies there in public?)

By the way, many years later, the guy who really started Neurodyne told
me that they actually intended to name the company "Cyberdyne," but
changed their minds after they saw the movie.

Last I checked, the uberchip and its use in missile defense are all gone
now, and the key guy doesn't even work
for the Air Force any more. Space-based missile defense is itself on a
very far back burner, apparently impossible for economic reasons if
nothing else. I used to think Sarah Connor reminded me a bit of my wife
in some ways... but really, my wife
is far more competent in military matters and training and sheer sanity
and willpower than Sarah Connor.. though I'm not sure how much that
helps. So far as I know, Neurodyne is defunct, with all its principals
scattered to the winds (though doing fine, no
one hurt).

And yet... this week... when I see very, very vividly and directly how
unlikely it is that humans will be able to get their act together on
issues like CO2 (due in great part to sheer mental deficiencies)... or
the threat that some of our cures just might be far worse than CO2
itself... while I see renewed powerful trends that could bring us to
even more powerful uberchips, and resurrect our technical capacity for
space-based missile defense... and all the cyber vulnerabilities
depicted in the final scene of Terminator 3 do seem quite real... as
missile defense people both in the US and China are fully aware of
one-on-one missile interception technology more advanced than what
Neurodyne intended to do with the robot arm (and more advanced than what
computatoinal neuroscience would dare to consider in most cases, with a
couple of possible exceptions in southern California)..
not taught or even hinted at in any computer science program anywhere so
far as I know... there is still some
reason to think that certain possibilities are not entirely off the table.

I don't know whether Dubai's nanoneuro chip program survived the drop in
oil prices. But cheer up, oil prices are well on course to come back...

Even so, my intuition tells me that new nuclear explosions killing lots
of people pose a bigger threat of
human extinction.

------------------------------

Anyway, enough old stories. If you don't believe it... well, you can
understand why I hesitate to consider
throwing in less believeable stories. What good would it do? I wonder.

Best of luck,

Paul

2 comments:

  1. Write your stories as just that, collected like _ This Way for the Gas, Ladies and Gentlemen_ or _ Kolyma Tales _ (the two 20th-century classics of totalitarian studies' "camp literature") . . . .

    Leave it up to people to decide if they're true, or how true, or not.

    ReplyDelete