|
|||||||
|
|
|
|||||
|
|
|||||||
From as early as I can remember, the idea that a computer,
or a robot could be sophisticated enough to operate like an autonomous human
being, was always interesting to me. In fact, when I was child, the sight of a Dalek really scared me, until I found out there was a
creature inside, in which case they were more like tanks than robots. But when I was a child machines with intelligence were only real on film
and radio, and it was great fun to imagine them talking and thinking, just like
us.
When the first computers began to sell to the ordinary
public, I saw the first Commodore 64 game, in which a small block on a screen
bounced between white paddles. With breathtaking speed PCs increased in power
and range and soon the first games came out which had layers of programing, which enabled them to record the moves and
choices of the players, and automatically make adjustments. Was this
intelligence? It seemed like it. But the problem was,
the computer could only make adjustments within the parameters of its program.
It could never generate new lines of code, and increase its efficiency beyond
what it was designed to do. The player, on the other hand, could continue to
increase moves and strategies, improvise, and innovate. The player was therefore
the intelligent one and the machine just a computing device. It seemed
intelligent, but it was an illusion.
Not long ago a computer called Deep Blue was designed to
beat the greatest chess player (Kasparov) in the world, and eventually it did,
but was this an example of intelligence? To understand what happened we have to
understand how a computer works.
Suppose I asked you what the Prime Minister had for
breakfast. Instantly you would say you did not know. Somehow your brain instantly
knows what it doesn’t know. Now suppose I asked a computer. It would begin a
search of every folder and file in its entire memory, checking to see if it had
the information. Eventually it would say it did not know, or “information not
available” but it would not then be aware of the fact that it did not know
something. That is one of the largest difference differences between human and
machine ‘intelligence’ - awareness.
To play chess you need to think perhaps one or two moves
ahead, so you must have a general idea of where all the pieces might be in the
future. A human can make a general scan, plan a strategy, and hope the other
player makes certain moves. The human assessment of chess is sweeping and
general, with a few sharply focussed plans. The computer on the
other hand, has to compute where each piece is, and where each piece might move
to, and if it did, what moves all the other pieces might make. This involves
millions and millions of possibilities, yet despite the computer’s mind-boggling
number-crunching Kasparov was able to beat the computer. He had learned how the
machine worked, and in the games he won or drew he was able to counter-attack.
The machine was unable to alter its method. All it could do was calculate every possible move. Kasparov was free to learn.
There are people who claim that “machines can think” but by
the word “think” they do not mean the same thing as when we say humans “think”.
True, there are cars about with onboard computers which learn the driving style
of their owners, and airline computers which check the progress of a flight and
help pilots make sensible choices, but these computers are simply following
machine code – lines and lines of commands – and they are not aware of what
they are doing.
In 1969 a robot was built which seemed to do many of the
things we would associate with intelligence. It was called ‘Shakey’
and it looked like a box on wheels, with some extra things on top. It could
roll about on a flat floor, view things with a camera, stop and start and turn
if it met with an obstacle, and a few other things. It was linked by radio to a
larger computer, which worked out what it should do, and it moved slowly. It
seemed to be behaving intelligently. Just as a human would stop and take some
other path to avoid an obstacle, Shakey could do the
same thing. It seemed to be thinking about where it was going, but it really
had no idea. It was just following machine code rules and doing what it was
designed to do. It didn’t even register as ‘intelligent’.
Isaac Asimov, a prolific writer of science fiction, came up
with ‘The three laws of Robotics’, but in order for any of these laws to
operate, a robot must be aware of many things, including itself, the nature of
the word ‘human’, the meaning of the word ‘injure’ and so on.
1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given to it by human beings, except where such orders would conflict with the first law.
3. A robot must protect its own existence, as long as this does
not conflict with the first two laws.
Let us suppose a robot is designed which can instantly recognise a human. This means it must be able to tell the
difference between a shop-window mannikin, a holographic
projection, a virtual reality picture, a dead human and many other variations.
It must also know whether the human is sane enough to give orders, or whether
the human is truly or just apparently in the way of harm. Humans frequently
indulge in very dangerous and risky sports, so a robot must be able to refrain
itself at the right times, which means it must have an enormous awareness of
human behaviour. In just this area alone the first rule is
enormously complicated.
The second rule is just as difficult, and the third adds
another layer to the complexity. If robots are ever designed to follow the
three rules they will need to be far smarter than humans at discerning the
correct actions to take.
For example, take a boxing match. Two humans
punching each other’s faces. A robot is not allowed, through
inaction, to allow humans to come to harm? How do you explain ‘sport’ to a
robot? There are vast variations of sport, from non-contact to high-contact.
Would robots constantly jump in to prevent “injury” – as they perceived it?
They could become the most hated nuisances on the planet! Every time a sport
began robots would be racing on to the field to stop it. Like cell phones
interrupting at a concert, they would have to be switched off, or updated
continually.
Which raises the whole question of what exactly is
“intelligence” because without a clear definition we might not be able to say
whether a robot is truly intelligent or not. As the joke goes, “Is there
intelligent life on Earth?” When we see some of the foolish things humans do,
we often wonder if they are indeed intelligent, but of course the joke is
cleverly biased towards the one who asks the question, assuming that we who
observe the Earth are the intelligent ones. Conversely, we might ask: is there
intelligent life in the machine?
A definition of intelligence is not too difficult.
But it might be useful to first define what is NOT
intelligent. Starting at the very bottom of the scale, when we see how gravity pulls
a dropped stone ‘down’, or towards the centre of the Earth, we know this is not
the act of an intelligent stone, but the law of gravity. The stone does not
‘know’ it is moving anywhere and the Earth is not aware of the stone’s movement.
Moving up the scale a little we see that many other laws of
Nature operate, and in some ways they appear to be doing intelligent things but
of course they are not. Take as one example among millions, the way water
freezes, flows and rises as a gas, take the way it forms rain, supplies all
living things with a vital ingredient necessary for life, absorbs heat and
distributes it around the planet, and so on. Water appears to be intelligent,
but we know it is just two atoms combined, and has no ‘brain’ or awareness of its
own existence. Yet it behaves intelligently, as if it was designed to do what
it does. This aspect leads us away from the water itself to some intelligent
designer of the water, but we must not follow that road at the moment.
A viruse seems to have intelligence. It
invades an organism, breaks into cells, pulls the inner parts of the cell to
bits and uses the pieces to construct copies of itself. But a virus without a
cell is just an exquisite but lifeless machine. Bacterium on the other hand can
move, react to their environment, reproduce, and form colonies that work
together as a whole – like ants or termites. Is this intelligence? Yes, based
on a very thin definition, but it seems to be blind intelligence, in the same
sense as a computer does what it is told to do. Bacterium
never do anything radical, creative, or ingenious. They cannot
think for themselves. They do what they have been programmed to do. They don’t
‘know’ they are bacterium, but they behave in vaguely intelligent ways, like
miniature machines, with a built in code for behaviour,
which they keep to strictly and never go beyond.
Insects are on a higher level of complexity, but again, they
seem to be not much better that bacterium. They have vastly complex lives, and
can behave in some innovative ways, but again, they are more like exquisite
robots than self-aware creatures – like bacterium only more complex. They do
what they are designed to do. They have brains, and they can make decisions,
and learn, but these features are part of their programming, and are often very
useful for their survival, and not the result of free will. Just as ‘Shakey’ could avoid an obstacle, an insect can shy away
from a wall, a net or a swinging hand. The response is instinctive, which means
pre-programmed, not an act of intelligence in the sense humans would react. The
contrast is a soldier who deliberately runs towards a machine gun, and
willingly lays down his life in his desire to achieve an ‘ideal’. An insect
would never give its life out of patriotism, or love of wife and family. Humans
frequently sacrifice themselves for ideals, but insects die in their billions
without ever considering what they are doing.
Moving up into the higher orders of life we come to mammals,
which are known for many traits similar to humans. For example Greyfriar’s Bobby, a small dog which stayed beside his
master’s grave for years until he too died, and there are many such stories, of
animals showing devotion and faithfulness to humans, as if they have similar
altruistic traits. Perhaps they do, and perhaps we underestimate the
intelligence of such animals, or perhaps they are simply following instincts,
which amount to very complex programming? Perhaps we attribute human traits to
animals because we are human, and that is the appropriate way to interpret
things. If we were machines we might attribute machine traits to animals.
The other word we need to define is “artificial”. When we
use the expression “artificial intelligence” we mean ‘intelligence which humans
have created’. This is very interesting, when you think about it, because (1)
humans naturally assume they are intelligent, (2) anything humans make which
acts in a way they think is intelligent is called “artificial” – as opposed to
“real” as in themselves. So humans have real
intelligence? If humans really could create intelligence, they might be
surprised if it turned on them one day and asked “But where did your
intelligence come from?”
If we are followers of
The machine may ask us one day “You gave me self-awareness,
but who gave you self-awareness?” In all the realm of Nature there are several
rules or laws which are never broken. Crudely put, one is the indestructibility
of energy, another is the direction of decay, and another is the fact that
nothing produces an effect greater than itself. In all life complexity always
stays either level, or goes downwards. DNA may lose small bits from its double
helix, but it never grows more complex. Living things may gather mutations and
degenerate, but they never improve – on the biomolecular
level. Any so-called exceptions are illusions. (For example the farmer who
breeds bigger cattle is simply sorting out genes for size, but losing genes for
other qualities. The orchid-grower may produce a flower of extraordinary
fragrance, but genes for other things will be lost in the process.)
If humans produce an inferior copy of themselves, an
intelligent robot, surely it is reasonable and logical to suppose that humans
were themselves the product of an even higher, and more intelligent creator, a
Being of immense intelligence, who knew how to design
bio-electric-organic-carbon
There are people who drag the definition of “intelligence”
down to its simplest and most basic meaning, but I suspect they do this because
they think human intelligence is only a more complex form of the simplest. In
other words, if you stuck a thousand insect brains together, you might come up
with a brain of greater intelligence.
In practise this never happens. Take
Deep Blue, the super computer programmed to play chess. Despite its enormous
computing power, it never asked anyone who it was, or why it was there. It
never got bored with chess. There are many other supercomputers which deal with
mind-boggling calculations, called mega-flops, but they never express a desire
to do anything but what they are told to do. Greater complexity in computers
never produces intelligence. If this was so, the global telegraph system, or
the Internet would have created a machine
intelligence by now. No amount of super-computers, all plugged into each other,
will ever produce self-awareness – which is part of intelligence.
In recent years scientists have been extending the abilities
of robots. Some machines can pick up sounds, view with cameras, detect odours, and even sense surfaces and
objects. In the crudest terms, robots can see, hear, smell and touch. In some
factories robot arms work using an ‘expert system’ program – in other words,
the robot is taught a repetitive task by a human. Assembly robots at car plants
are taught, or led through their duties, and each move they make is recorded by
a computer. After some practise the robot can then get on
with the job unattended, using its program, but this is not intelligence. The
robot is not aware of what it is doing. It has about the same hope of
questioning the factry manager about its appearance
as an electric toaster has of asking for brown instead of white bread.
Robot explorers have been and are being designed to trundle
across other planets. For this reason they are ideal, because they can explore
places where it is too dangerous for humans to go. They can crawl into
volcanoes, chug across the sea floor, stare into the insides of bombs and creep
down tunnels. In 1970 the ‘Mobot’ was designed to handle
dangerous materials – it had grippers, cameras and a crane. Humans operated it
by remote control.
Many scientists have thought how handy artificial
intelligence might be for a space probe. A machine with AI has a better chance
of survival than one without because there can be a long time-delay in
communications from Earth. Even travelling at
186,000 miles per second (300,000 km/sec) radio waves may take hours to reach a
probe explring a far planet. Disaster would strike before a
command could be sent from Earth. Future robots may have to look after
themselves when in trouble, so programs are being designed to give the machines
some “common sense”.
But once again there are the two problems. First, any robot
designed by man which has AI begs the question as to where man himself got his
intelligence, and secondly, if man can program ‘common sense’ into a machine,
all the machine can do is follow a predesigned set of
rules. It only appears to have common sense – it is virtual
intelligence, not real intelligence.
Which brings us to Australian robot
scientist Rodney Brooks, who likes to build robots. One of his creations is called “Cog” after the Latin word
‘cogito” meaning “I think”. Cog does not look much like a sci-fi robot. He is a
screwed-together collection of wires, tubes, nuts and bolts. He does have eyes
and a body, which is basically similar in shape to a human. His eyes are
cameras, his ears are microphones, and he has sensors in his arm which prevent
him from breaking things.
Mr. Brooks uses a form of programming called ‘layering’. For
example, if a robot meets with an obstacle it follows a line of instructions
which tell it to ‘step over’. But if the object is too large it follows another
line of code that says to ‘go around’. If it cannot ‘go around’ another line of
code says ‘back up and try from another angle’. Layer after layer is tried
until the obstacle is either passed or the robot stops, or possibly it tries
another direction – but this is not thinking, in the human sense. This is a
machine following commands. A human might decide the whole idea of going round
the obstacle is silly, or it might jump against the obstacle and make up a
game, or paint the obstacle, or blow it up, or compose a piece of music about
it. Humans are vastly superior to machines operating on machine code.
But what if a robot, that is, a computer, was designed which
could gradually ‘learn’? Here again we have a problem, because it is not just
the amassing of information which makes humans intelligent. There is more to
human intelligence than merely storing data. If this was all it involved, we
would expect to find that computers which stored the most data were the closest
to being intelligent – but of course they are not. There are super-computers
which hold fantastic amounts of data but they are no different than a pocket
calculator when it comes to intelligence.
The Internet is like a global computer, which has
immeasurable amounts of data available, but it still operates like a simple
library book retrieval system. It may find 40 million entries for a single
subject, but it never finds self-awareness. Logically, if all the knowledge in
the world were stuffed into a human brain, it would not make a human more or
less human.
Another branch of robotics uses the idea of neural networks
to build a ‘brain’ for the machine. These networks are based on the idea that
if computers are built on the same lines as the human brain, a similar result
may follow. In the human brain billions of cells, called neurons, are packed
together in an amazingly complex system. The neurons receive signals as electric
pulses from all over the body, and they process this information in a complex
sort of way. In some ways each neuron seems to ‘decide’ the importance of the
impulse it receives, and on the scale of billions this is an organisational miracle.
The brain runs two basic systems. One is for the function
and maintenance of the body, and the other is for self-awareness, and
decision-making. These two are often called ‘conscious’ and ‘unconscious’. Let
us pause here and think about these two for a moment.
The unconscious function of the brain is so vast in its
effects that science has only just started to work it out. From the immune
system, to the growth and maintenance of every cell, to the reception of input
from every sense, through to evaluation of data and ‘filtering’ of input from
the senses, healing, blood pressure, and even the bumps and ridges on the
fingertips . . . and all this is done automatically, without us having to
‘think’ about it! All the computers in the world could not manage this amount
of processing for just one human, yet the human brain manages it every minute
of our lives without us even being aware of it.
The conscious function of the brain is like a voice, which
we can actually hear, in a mysterious ‘inward’ way, inside our head. As we
consider things, discussing them inwardly in our own language, we may also view
what we are thinking about on an inner screen, and imagine possible outcomes,
and improvise. We can also respond to situations in an ethical or moral way –
unlike all other creatures. We talk about something being either right or
wrong, and these norms are generally agreed to by all other humans, and have
been agreed to by all cultures at all times in recorded history. (See the final
notes in C.S.Lewis’ ‘Abolition of Man’)
Animals, such as dogs, cats, chimps and elephants have been known to show signs
of sadness when one of their herd dies, but in most cases animals
carry on with life regardless of what dies beside them. Humans bewail the loss
of loved ones, memorialise dead soldiers, set anniversaries for dead heroes, but animals usually
just carry on living.
The conscious part of human thinking raises us above all
other living things. We know that we know something, and then we know that we
know that we know. While a machine may instantly recognise
a face, it never feels anything. A computer never becomes aware of anything, in
the same way humans do. A machine may be built which breathes using lungs, but
the machine will never know the pleasure of breathing.
Neural networking is a grand idea, but in practise it always fails, because humans do not fully
understand how the brain works, and the neuron structure is too complex to
copy. As well as this, the connections, sometimes 10,000 per neuron, means
there is an almost infinite number of possible pathways available for the
brain, whereas human engineering cannot come even close. Even the best AI
machine is not much better than an insect, and even if this could be improved
on, the step from insect to human is not just a matter of multiplying the
circuitry.
Just pause for a moment to consider the human brain. It contains about 100 billion neurons. (This would be about the same as one million bee brains stuck together.) Each neuron is interconnected with all the neurons all round it, in a three-dimensional structure. When humans make computer curcuits, they add fixed connections, but the human brain connections are always changing – new connections are always being made, as we learn new skills and store new data, and at the same time some 200 different chemicals wash through, altering things every moment of every day and night. The power of the brain is huge, and its processing speed faster than any computer yet invented by man. Such amazing design features always run ahead of Man, urging him to learn more and to understand more, which means that we have a long way to go before true ‘artificial intelligence’ is produced by intelligent Man.
Many predictions of the modern future include robots, or cyborgs, (from ‘cybernetic organism’) which is supposed to
be a mixture of man and machine. True cyborgs
do not exist yet, but the idea has been around a long time. People have tried
many times to imitate Nature by adding mechanical or electrical parts to
themselves, but this only reveals that no matter how clever Man may be, “Nature
thought of it first.” Some of the cyborg ideas
include mechanical hands, arms and legs, computers linked to human brains,
brain implants that include light sensors, telephones and even the Internet!
Experiments at the Max Plancke Institute have shown that joinng living nerves to microchips is possible – in rat’s
brains at least – so it may be possible to enhance human abilities one day with
electronic eyes, or even boost intelligence, or replace damaged brain cells, or
download memories on to a disk.
But the future may also turn out to be very similar to the
present, and all the hopes and dreams of scientists to create AI may fall flat.
Just as time-travel, and prevention of ageing, and abolishing death, and
the eradication of crime are no nearer to achievement than they were thousands
of years ago, so true intelligence in machines may never be realised. The main reason why I think this, is because “intelligence” in the human sense is not
just a matter of processing data. Humans do not run on machine code. They are
several levels higher than machines, and a simple comparison between the best
possible AI machine and a human reveals this.
A true AI machine would have to be aware of itself. It would have to be able to communicate on its own
volition, which means it would have to operate independently of its
instructions. It would also have to want things, rather than need them, for
example, it might want to be left switched on rather than off, or it might want
to ask a question about something which was not directly relevant to its work.
Humans of course do all these things as a matter of course.
The origin of machines is humans. No machine ever evolved.
Just as Man looks downwards at his machines, and sees the
work of his hands, why cannot Man look upwards and see that he is the work of
someone Else’s Hands? Of course we are speaking here of God.
Consider these Bible verses:
“In the beginning God created the heaven and the earth.”
Genesis 1:1 (Here we identify the designer of the environment.)
“And God created
great whales, and every living creature that moves, which the waters brought
forth abundantly, after their kind, and every winged fowl after his kind: and
God saw that it was good.” Genesis
“So God created man in his own image, in the image of God created he
him; male and female created he them.” Genesis
Just as God created Man in His own image, Man is now
trying create machines in his own image. Just as the
greater produces the inferior, Man the inferior to God is now creating machines
inferior to himself. We can safely predict that Man will never make anything
superior to himself, because a living human is too vast, too complex, and
contains too much built-in design-wisdom for mere humans to replicate, let
alone improve on. But in the same way, Man needs a lot of improvement too.
But what if the world we know today was not like the
original world first created by God? That would change things. What if God
originally created a perfect world, in which there was no sickness or death, no
violent weather, and no carnivors, but subsequent to that
Nature began to change? This is precisely what the Bible does say.
When Adam and Eve, the first created humans sinned, they
brought with their sin a punishment into the world, which had a sort of gradual
down-grading effect on Nature. Death entered the world, along with ageing and
sickness. Genetic mutations began to appear, species became extinct, Nature
became badly balanced and the ecology of the planet began to fall apart. The
environment changed. Weather systems appeared which struck the planet with
force. That is how things have been since sin entered, and that is how things
will be until God calls an end to the age.
In the meantime, consider what God has done to help. From
the very first sin God provided a substitute for sin, a sacrifice. The animal
died in place of Adam and Eve so that, when they trusted in the sacrice for them, God forgave them and gave them a door
into His new world – which He would one day open to them. Until then they would
have to sleep until the day of resurrection.
Down through the ages God allowed people to make sacrifices for sin, but finally God gave His only Son as the one final sacrifice. This was the turning point in world history. When Jesus died on the cross, He revealed to the world the character and love of God. The sinless Saviour took on Himself all the sin of the world, and died for us, while He Himself never sinned. Such was the Saviour’s love.
It may seem that we have changed to another topic, but we
have not strayed far from the subject of articicial
intelligence, because the theme of this essay was intelligence, and it takes
intelligence of a certain kind to understand the gospel. It is utterly
impossible to explain in words what God has done - to an animal. Even the most
intelligent parrot, or the smartest chimpanzee, would never come close
to understanding even one simple verse from the Bible. But humans can
understand, because they are made in a way that is similar to the God who made
them. God uses language, so do humans. God uses logic, so do humans. God loves,
so do humans. God never sins, but humans know what sin is, and they know when
something is sinful.
The Ten Commandments define sin, and Man understands them
because God has also written them on Man’s heart. The communication between God
and Man is written in a book, the Bible, and expressed in every spoken language
of the world. Man cannot communicate with machines, except by writing lines of
code. Machines will never speak back with love, or show genuine appreciation,
and machines will never sin, or hate, or rebel. They will never have the
capacity to do any more than what their program allows.
Man on the other hand is a freewill agent, unlike the
animals that obey instincts from birth through to death. Man can say yes or no
to orders. Man can accept or reject an offer, even when it in his best
interests to accept. Man does not have AI. Man is not a machine. Man may share
a few things in common with animals, on the basic ‘animal’ level, but after
that, Man rises far above animals, because Man has an aspect of intelligence
that no animal shares: self-awareness and conscience.
“For thus says the LORD that created the heavens; God himself that
formed the earth and made it; he has established it, he created it not in vain,
he formed it to be inhabited: I am the LORD; and there is
none else.” Isaiah 45:18