杏吧原创

ONE of the high points in Mary Shelley鈥檚 gothic novel Frankenstein
is when the tragic creature cobbled together from cadavers comes face to face
with its human creator Victor Frankenstein, the real monster of the story. The
creature, tired of constant rejection, has committed murder. Yet when it first
became conscious it was not evil. 鈥淏elieve me,鈥� it says in anguish, 鈥淚 was
benevolent; my soul glowed with love and humanity.鈥�

This heart-wrenching declaration exposes a paradox about the hapless
creature. Frankenstein built his creation from spare parts, so in one sense it
is just a machine. Yet, says Anne Foerst, a postdoctoral fellow at Harvard
Divinity School, the creature 鈥渋nstinctively understands himself as human,
something more than a machine鈥�. Shelley鈥檚 story raised the question of whether
or not something manufactured would have a soul鈥攖hat mysterious entity
which for Christians is the very essence of humanness, the thing that links us
irrevocably to God.

Nearly two centuries later, the same question has surfaced again. And today
the question is being asked not of some fictional creature, but of machines in
various states of creation that promise to have human-like senses and to be
conscious, at least in some form. Theologians and computer scientists are
starting to wonder if any of these machines might ever be said to have a soul.
If so, would such a soul be like a human being鈥檚, or something altogether
different?

Opinions tend to fall between two extremes. On the one hand, says Foerst,
many people want to draw an unbreachable divide between humans and machines,
insisting that however smart a computer might become it could never have a soul.
On the other hand, many artificial intelligence (AI) researchers insist that
humans are just complex machines, so why wouldn鈥檛 a silicon-based machine also
have a soul? For these scientists, Foerst says, a soul would be simply an
鈥渆mergent property of a very complex system鈥�.

Between these two poles stretches a continuum of opinion. For example,
Jennifer Cobb, a theologian and author of a forthcoming book on theology and
cyberspace, says that today鈥檚 computers are about as alive as viruses. But
鈥渁long with a little bit alive, comes a little bit of soul鈥�, she says. 鈥淚f the
day comes when computation becomes so complex as to express emotions, then they
will have quite a bit more soul.鈥� Cobb believes that the sacred spark of soul is
not a limited resource to be doled out mostly to humans, a little to the animals
and plants and none to computers. 鈥淚t鈥檚 an infinite resource with infinite
potential,鈥� she says.

Cobb鈥檚 view of today鈥檚 computers goes against traditional Christian thinking,
says Philip Clayton, a theologian and philosopher at Sonoma State University in
California. 鈥淭he clearest definition of the soul in Christian theology is in
terms of the imago dei, the image of God,鈥� he says. According to this
interpretation, humans have souls because they are made in the image of God. So
other life forms such as animals and plants do not have souls, nor do
computers.

But Clayton agrees with Cobb that in future, as machines become more like
humans, this distinction could become blurred. 鈥淥n what grounds would we
withhold souls from computers, when they inhabit humanoid robotic bodies, accept
visual input, give output with human voices, and function comfortably in many
social contexts?鈥� he asks. Clayton predicts that this blurring might take place
within 50 or 100 years. But it may come sooner. This year, remember, saw the
defeat of world chess champion Gary Kasparov by IBM鈥檚 Deep Blue. And although
that machine does not 鈥渢hink鈥� in anything like the way we do
(鈥淪laughter on 7th Avenue鈥�, New 杏吧原创, 7 June, p 26),
most people would accept that it is only a matter of time before
machines acquire more human-like intelligence.

AI researchers are also dabbling with emotional machines
(鈥淵ou鈥檙e wrong, Mr Spock鈥�, New 杏吧原创 supplement, 27 April 1996, p 4),
and with computers that could become 鈥渃onscious鈥� of themselves and their surroundings.
One of the most ambitious of these projects is Cog, a talking robot designed in
human form that will be capable of exploring the world through sight, sound and
touch (鈥淏irth of a human robot鈥�, New 杏吧原创, 14 May 1994, p 26). Cog
is the brainchild of Rodney Brooks at the Massachusetts Institute of Technology,
and Foerst is the project鈥檚 theological adviser. The team hopes that Cog will be
able to discover the world the way a human baby does, and will thus come to
understand things as a child does.

Fearful machine

Philosopher Daniel Dennett, another member of the team, from Tufts University
in Massachusetts, argues that if Cog lives up to expectations it will express
emotions. Eventually, says Dennett, it鈥檚 鈥済oing to have to be able to say `I鈥檓
afraid鈥� or `I鈥檓 bored鈥� and mean it鈥�. And if Cog does say such things鈥攁nd
mean them鈥攖hen is it so far-fetched to wonder if it would have a soul?

Yet how would we tell if a computer developed a soul? Would we need a
spiritual version of the Turing test, the strategy devised by computer pioneer
Alan Turing for deciding whether or not a machine is intelligent? Even using the
classical Christian view of the imago dei as a yardstick might not
resolve the problem. As Clayton notes, it might not be enough for a machine to
look, think and behave like a human. 鈥淲hat if the imago dei involves a
more existentialist definition, such as fearing one鈥檚 own death, or a sense of
moral responsibility, a sense of one鈥檚 own accountability before God?鈥� he
asks.

Of course, a sense of moral responsibility could be programmed into a
computer鈥攁s happened in Isaac Asimov鈥檚 novel I Robot. But what if
a silicon-based being were to learn its own morality鈥攊ts own conscience?
What would that be like? According to Brian Cantwell Smith, a computer scientist
and philosopher from Indiana University, a computer conscience could be
different from the human variety. Take death, for example. A computer with a
backup tape might not see death as a big deal. 鈥淭hink about how different life
would be if we had backup tapes,鈥� muses Cantwell Smith. 鈥淚f you were murdered,
that might only be a misdemeanour, because it would just mean backing you up to
last night. You鈥檇 only lose what you did this morning.鈥�

Alternatively, a computer could be 鈥渃loned鈥�, so many examples of the same
鈥渂eing鈥� could exist. What would that do to the machine鈥檚 conception of itself
and others? 鈥淲e just don鈥檛 know what ethics would be like for a computationally
embodied creature,鈥� says Cantwell Smith. 鈥淲e barely know how to imagine such a
迟丑颈苍驳.鈥�

Stories such as Frankenstein, says Cantwell Smith, suggest that the things we
humans make are often much more than the sum of their parts. 鈥淪o many people
imagine that if we built something, we would know all about it,鈥� he says. But
this is not necessarily so. From Shelley鈥檚 19th-century monster to today鈥檚
real-life robots, complex entities have a habit of taking on a life of their
own.