If you can't breed your supermen and you can't create them by defying
the laws of man and God, then you can always fall back on the old
standby and build a mechanical replacement for mankind; i.e. androids.
As everyone knows, androids are a cut above human beings; they are
inherently smarter, stronger, faster, and (generally) morally superior
to their creators and are damn smug about it. But
what "everyone knows" probably has more to do with hubris than
feasibility. At first, creating a android seems a doddle.
After all, the human body is just a machine made out of meat.
But as soon as you start looking at the problem it turns out that
humans are incredible pieces of engineering and trying to create a
mechanism to human specifications is a something that we are
still way beyond even remotely accomplishing.
Take a look at the
Yul Brynner android from the film Westworld, for example.
Never mind all the relatively easy tasks such as drinking, shooting
a rifle, or the handling the incredible minutiae of appearance,
subtleties of expression, or even walking down a dirt road.
can ride a horse, climb a ladder, and look convincingly human even
while reeling backwards from a hail of bullets.
Then there is the vexing paradox of what happens when a machine is
endowed with something approaching human intelligence (we'll leave
aside if it is even theoretically possible to equal or surpass it). Our
machines appear impressive because we can suit the tasks they do to
fit their inherent limitations; otherwise they'd come to a dead stop
at the first blind alley. Bowling pin machines don't need to
play chess and chess computers don't need to set up bowling pins, but
a human being is expected to be able to do both, though not at the
Another difference with machines is that they cannot understand an unintended
opportunity when they see it. We can make a machine to sell chocolates at a
railway platform, but we can't make one that will walk up to someone,
sell him a Cadbury bar, overcharge him, trouser the difference
for itself, and then lie to the home office about the whole thing to
cover its backside. Human beings do this sort of thing.
Humans also have to deal with all the
ambiguities and unknowns of the real world. They have to act
whether they are certain of the outcome or not and make the best of
the results. Therefore, they make
mistakes no matter how clever, talented, or well educated they are.
If an android is going to get anywhere in the real world, it must be
able to muddle along on less than complete or even reliable data–maybe even none at all. That means that any machine that has to work on a
human level has to be expected to act with less than optimum results
in response to less than optimum data.
In other words, the more sophisticated the android, the more it's
going to screw up. And the more it approximates human
intelligence, the more it will make a right cock up of things. That raises the awkward question:
If androids are going to be as prone to failure as we are rather than
infallible supermen who can't use contractions, then what's the
point of building them in the in the first place?