Clever Hans in the Age of AI
by Broadbrush Brightley
Despite recent doubts by the likes of Calcikopff and Horsdvors, the
reality of AI, or artificial intelligence, is generally regarded to
be a teleological fait accompli: sooner or later, they argue, computers
will actually think. Of course, with the backing of many a member of
MIT's Media Lab and corporate IT departments anywhere, this conclusion
is unquestionable. I would here like to avoid downplaying the great
strides made in recent years by computer technologists in promoting
this evolutionary vision. But certain aspects of this technological
revolution have, indeed, already been achieved. Parallels between current
AI eschatology and the triumphs of one Clever Hans are striking both
in terms of those creating the technology and that which expresses it.
The story of Clever Hans is too famous to bear repeating in detail
here, but one should be reminded that Hans's ability to do simple addition
and subtraction was facilitated by his handler unconsciously giving
him signals about how many times he should move his hoof to indicate
the correct answer to the questions put to him. This behavior was reinforced
when the handler gave him food for correct answers. His equine intelligence
is hardly surprising just on the basis of the facts alone: similar miracles
have been developed to exploit the grand smarts of pigeons, chicks,
pigs - the what-have-you of the animal world - by many a diligent entrepreneur.
What was extraordinary about Hans's case was that the horse's handler
himself had no idea he was giving the subtle signals that led to the
miraculous results. If the answer was "five" Hans's handler
would inadvertently nod the correct number of times and Hans would follow
suit, clomping with his hoof five times, knowing full and well that
this game of "horsey-see-horsey-do" would result in a tasty
treat. The handler had spent so much time with Hans that a delicate
feedback loop had developed, increasing the horse's abilities and making
the handler himself part of the functioning mechanism; Hans and his
handler had become the first documented cybernetic system.
Unfortunately, the preponderance of zoologists and developmental psychologists
view the achievements of Clever Hans as a simple ruse. They accept the
scientifico-skeptical tradition that fails to see that, through his
unconscious conditioning, Hans and his handler became cybernetic. This
same scientifico-skeptical tradition has led to the other cultural trends
such as "irony," which, as Purdy implies, is quickly leading
to this civilization's downfall. Such anti-techno-triumphalist trends
are even more reason that AI researchers should move in an even quicker,
more practical way teleologically, as we shall see below.
Research by Squeamous et al has indicated that those intent on developing
AI tend to spend much time refining and tweaking their hardware and
software systems, generally well over 80 hours a week (14). They tend
to be almost fanatical in their pursuit of what they have predetermined
as possible ("Learnings" 144). From this develops an interactive
feedback loop, with AI systems and their creators in cybernetic fusion.
AI systems are designed after all to be learning machines, incorporating
new environments into their programming. Smedley writes:
On the first day, SPAX 109 was an idiot - unresponsive
to any but the most simple stimuli, mimicking my voice
patterns in the most simplistic way. After a rigorous
week of television viewing with a variety of channels
moving through his sensory apparatus in a random
fashion, SPAX was quoting everything from Britney
Spears to William Shakespeare with all the coherence
and proficiency of a college freshman.
Smedley adds this, realizing what he has wrought in the now famous
SPAX experiment: "[S]urely a life has been created." Just
as Clever Hans picked up on the cues of his handler, here the SPAX 109
Thinking Machine created by John Smedley of the Erstwhile Institute
of Technology in Farmingout, North Carolina picks up on his specific
environment and social milieu. An intelligence is created through the
interaction of the created and the creator, the evolvement of the beast
through the man.
Alexandra Busch-Snell demarcates another frontier of the AI realm.
Here computer modeling is a source of eternal life. After a complete
scan of neural pathways and functions, one's brain is modeled on a powerful
computer; all one's thoughts, experiences, memories, traits, are downloaded
and run, essentially, indefinitely (92). Similarly, in his comprehensive
history of the matter, Phillip Phillipi writes about a movement in the
state of Prussia after the phenomenon of Clever Hans to replace the
brains of healthy horses with the brains of dying men and women (282).
The concept was inspired by Hans's cleverness itself. The movement's
founders, Johann Husskopf and Gretchen Froelichingenger reasoned that,
given the horse's innate intelligence as demonstrated by Hans, the horse
body could be made the perfect vessel for the extension of our all-too-brief
human lives. The movement, which quickly became associated with the
mythological figure of the Centaur, was the first of several such movements
that galloped across western Europe the forty or so years following.
Thus the Alpha Centaurian movement gave rise to the current expression
"as healthy as a horse" to describe their own practical eschatology
(Phillipi 310-30).
The movement's popularity can be attested to by the Prince of Nockwurst
of Bavaria's bequest of 400,000 marks to the Alpha Centaurians to continue
their quest, while bankrolling the movement's research arm day-to-day
(Phillipi 742). It was a substantial investment in pure science, one
of the largest of the time, and, under the auspices of a Dr. Argent
Angstweiler Langweilich, great progress was made. In fact, the bequest
came into its full fruition when the Prince expired in an early, failed
experiment. The donor horse was fully tranquilized and his brain removed,
but the Prince's brain, sadly, met with an unplanned critical trauma
when it was accidently dropped into a bucket of oats (Phillipi 1020).
Had this or the other 27 attempts conducted by the Alpha Centaurians
succeded, no doubt 20th Century civilization would have been much better
for it. Certainly, it would have marked a tremendous increase in the
vitality of the hay and forage industries. Increased strength, vitality
and grace would also have ensued as the human population made the transition
into the Centaurian/cyborg model, and the need for such an automobile-intensive
culture could have been avoided entirely. A parallel decline in the
coach building industry would not, however, have been avoided.
Perhaps the most unfortunate failure of the current AI movement is
its lack of boldness. For every new university program or startup business
investigating the eternal-life aspects of AI, there are none actually
attempting it. One would hope that such diligence as contemporary AI
investigators apply would be followed up by a few practical applications
of what they know or have discovered. Pioneers in this field have much
to lose - perhaps all brain functions, perhaps more - but they have
much more to gain: the creation of an eternalized human race, unfettered
by the degradations of bodies and the imperfections of "RL."
Virtual smoking won't kill you nor give you emphysema, so the downloaded
being would be able to engage in as many virtual cigars as she likes.
She could drink as much and eat as much virtual red meat as she liked
when downloaded. Such existence would be pure bliss, the next logical
step in our evolution. Combined with the learning machine concept, the
downloaded self could instantaneously adapt to new virtual situations,
as shown in the prophecy The Matrix, creating, potentially, a
universe of ever-expanding minds.
So what is stopping the current researchers from more
directly pursuing the ultimate dream? Undoubtedly it is cowardice. AI
provides the next great frontier. With careful modeling after the old
one, learning from their new environments, with the downloading of current
intelligences we will be able to gain some kind of perfection. Who else
but us, now, should be the creators of the next world? After all, we've
done such a good job with this one.
Works Cited
Busch-Snell, Alexandra. "Temporal Cyberprojection Through Ultimate
End-User Modeling." Studies in Contemporary Biomechanics
44 (1998): 90-120.
"Learnings: a Cybervision." Proceedings in Cybernetics
4.1 (1999): 122-47.
Phillipi, Phillip. The Alpha Centaurians: a Comprehensive Study.
New Punswick: Knopfwurst, 2000.
Smedley, Joshingua. "It's Alive, By Me." Cyberjournal/Diss.
Erstwhile Institute of Technology, 2002.
Squeamous, Crampillion. "Wonks in Cubeville." Progressive
Work Studies 17 (2001): 10-22.
