Some computer-scientists are rewriting the creation myth to place computers at the pinnacle of creation. Will super-intelligent computers take over the world? Dr. Mae-Wan Ho takes a cool long look and discovers some surprising answers.
British playwright Alan Ayckbourn's Comic Potential opens with the scene of a young man lying in a hospital bed with the mother crying at bedside. The doctor is trying to tell the mother of an operation that has to be done to save the boy's life, and a nurse is standing in attendance. In the gloom above on a platform sits a man watching and directing the scene.
The doctor takes a clipboard from the nurse and shows it to the mother.
"There! You don't have to be an expert. Just look at these X-rays, woman. They speak for themselves. There's massive damage to the foot. I'm going to have to operate immediately. I'm going to remove the temporary pluster cust and umputate just above the unkle "
The director curses. Behind him, a flustered computer programmer looks in dismay at her console, and the technician tries to explain what has happened. The programmer and technician have been operating controls to make the 'actoids' - robot actors - do what the director wants.
The software has glitched, making the doctor substitute 'U's for 'A's. But what's the matter with Nurse? She isn't programmed to laugh, and certainly, isn't supposed to laugh at jokes.
Worse yet, love subsequently blossoms between actoid Nurse and human Adam, a visiting fan of the director's old movies, before the age of actoids. A robot that laughs at jokes and falls in love? Is she conscious like a human being?
Of course, no robot today can laugh at jokes or fall in love, no robot can even tie a pair of shoelaces. Decades of artificial intelligence research have yielded nothing that approaches natural intelligence.
That's not to say robots can't do some impressive routine tasks. A computer has beaten the world's chess champion by dint of calculating the consequences of all the possible moves many steps ahead in order to choose the best. Sophisticated computer programmes can write poems, make music and even paint pictures. Computer games, even for someone like me who hates them, are marvels of animation. I, for one, hardly ever write except on a computer, and where would any counter-culture or protest movement be without the internet? I am told that several recent revolutions were won with the help of the internet.
In a fascinating book, Technomanifestos (2002), Adam Brate charts the history of the digital age, from cybernetics to the World Wide Web, through the biographies of a collection of radical visionaries who brought it about. I was especially impressed with the attempts of people like Norbert Wiener, Vannevar Bush, J.C. R. Licklider, Doug Engelbart and Tim Berners-Lee to keep knowledge open and circulating, enabling information technology to thrive and develop in the public domain, even after the likes of Bill Gates have set up private monopolies. The now played-out IT boom would probably never have happened but for the collective efforts of computer scientists working in all corners of the world sharing knowledge and software without reserve and with little or no financial reward.
But more importantly, those pioneers shared a dream. "[T]he goal of the information revolutionaries", Brate tells us, "is to create new systems - technological, social, political, and economic - that adapt to people instead of the other way around."
What has happened to that dream since? It has all the signs of turning into a nightmare, so let's take a cool long look.
First, there's a contingent of computer scientists who are convinced that when computers become many times faster than it is now, and more complex, they will be more intelligent than human beings, and may become conscious, like human beings. They may even take over the world, for better or for worse.
That's what Ray Kurzweil, inventor of reading machines for the blind, music synthesizers and speech-recognition technology, predicts in his book, The Age of Spiritual Machines, published in 1999. He says that by the second decade of the present century, it will become increasingly difficult to distinguish between human and machine intelligence. And, as computers become more complex, "they, too, will necessarily utilize goals with implicit values and emotions".
Is there really no difference between human and computer intelligence?
"Are computers thinking, or are they just calculating? Conversely, are human beings thinking, or are they just calculating?" he asks, and answers, "The human brain presumably follows the laws of physics, so it must be a machine, albeit a very complex one."
Furthermore, if a person scans his brain through a non-invasive scanning technology like magnetic resonance imaging, and downloads his mind to his personal computer, "Is the "person" who emerges in the machine the same consciousness as the person who was scanned?" Kurzweil doesn't say yes, but invites his readers to think that may be the case.
Better yet. The downloaded 'person' may find herself with a perfect, synthetic, 'nanoengineered' non-biodegradable body, thereby achieving immortality. Sounds tempting, doesn't it, never more to suffer the humiliations of the flesh, like growing old. We could keep on down-loading our 'person' to new, better-styled models as the one before goes out of fashion. "We will be software, not hardware." Kurzweil says, reminding us to constantly back ourselves up against a computer crash.
Kurzweil believes that once computers become faster and sufficiently complex, they too, will acquire a personality, if not a soul, rather like the one we downloaded on the computer. But won't that machine end up with a split personality? One computer-emerged, the other human-derived?
Well, that's where the 'spiritual' machine comes on the stage. "We don't always need real bodies." Virtual reality proves as much, Kurzweil reminds us. "Virtual reality is not a (virtual) place you need go to alone. You can interact with your friends there (who would be in other virtual reality booths, which may be geographically remote). You will have plenty of simulated companions to choose from as well."
You've guessed; virtual sex is next, even virtual nirvana.
I must confess I was fascinated and revolted in turn by what Kurzweil has in store for us. It took me two years before I felt I could respond.
Kurzweil glosses over some large, unjustified and unjustifiable assumptions. The first is that consciousness will automatically emerge past a certain threshold of complexity, however complexity is defined. So although computers today are not conscious, those made in a couple of decades from now might well be.
There is an equivalent assumption in biology that lowly organisms aren't conscious, and only organisms above a certain evolutionary grade could be considered conscious, which is completely arbitrary. To me, all organisms are conscious, however 'lowly', simply because they all grow and develop autonomously and independently of us, and more importantly, they each have their own 'purposiveness' in life that cannot be subverted in an arbitrary way. If my computer one day should satisfy those criteria, then, and only then would I regard it as conscious, and treat it accordingly, and yes, could even fall in love with it. I don't see that as a big problem.
The second large assumption Kurzweil makes is that the human brain is not only a 'machine' - subject to the laws of physics - but a machine of the same kind as a computer.
Invoking the 'laws of physics' isn't much, because the 'laws' have changed and will change again as we find out more about the strangeness of matter and energy. It is just shorthand for 'not by special creation'. But even if the human brain is subject to the laws of physics, it could still be a very different kind of machine from a computer. The organism, including the brain, is indeed a very special kind of 'machine'.
In my book, The Rainbow and the Worm, The Physics of Organisms (1998) and elsewhere, I have described the organism as an autonomous, self-sustaining domain that stores and mobilises coherent energy over all space-time scales. In the ideal that approaches quantum coherence, every single part, down to the individual molecule, is so perfectly correlated with the whole that it is also maximally free. What makes it possible for the organism to achieve this state is the liquid crystalline continuum - the stuff of flesh and blood - that makes up its body (see box).
Kurzweil is also assuming that the human brain is identical to 'mind', which many philosophers will dispute. To me, it is like saying that music is nothing more than the amplitudes and frequencies of all the sounds made in succession, or that a painting is the rgb (red, green and blue) values of each pixel on the computer screen.
Then there are the related, subsidiary questions. Is 'mind' the same as a 'person'? Will a 'person' really emerge in the machine? Is the body not part of the person?
There's one thing a 'person' is not. She's not a static collection of software programmes stored in the hardware - the brain - whose function is to 'process information'.
The information processing metaphor has taken over biology completely. Kurzweil is following Francis Crick, who first introduced the language of information technology into molecular biology, and gave rise to the most hard line genetic determinism - the idea that DNA determines everything that goes on in the development and behaviour of living organisms. Not coincidentally, Crick has since gone to work on 'consciousness', which he supposes to be "the activities of nerve cells in the brain". His book, The Astonishing Hypothesis (1994) is a major contribution to the 'mind as computer' ideology.
It is interesting how the same set of ideas is bounced back and forth among the major disciplines in the mainstream, in an never-ending cycle of mutual reinforcement that becomes utterly impervious to invasion by outsiders or, more seriously, to contrary empirical findings.
Not just mind, but the whole of evolution is seen as computation.
"Evolution is a master programmer." Kurzweil writes, "The software programs have been all written down, recorded as digital data in the chemical structure of an ingenious molecule called deoxyribonucleic acid, or DNA The master "read only" memory controls the vast machinery of life."
"The DNA code controls the salient details of the construction of every cell in the organism, including the shapes and processes of the cell, and of the organs comprised of the cells The synchronized flexing of muscle cells, the intricate biochemical interactions in our blood, the structure and functioning of our brains, and all of the other diverse functions of the Earth's creatures are programmed in this efficient code."
And how has all that come about? By the natural selection of random variations, which can be perfectly simulated on the computer. That's what evolutionary algorithms are all about. You let different algorithms compete and the best wins. But no evolutionary algorithm has yet produced life. Nor has the much-touted random association neural network research produced anything like a robot that learns as the human infant or an animal learns.
James Watson seduced the public into spending billions on the human genome project by promising to reveal 'the blueprint for making a human being'. But the myth of genetic determinism has been well and truly exploded as the human genome sequence was announced by chief sequencer Craig Venter, who has now gone on to run a not-for-profit organisation, as his pronouncement was definitely bad for business (see "Biotech fever burning, burning out" I-SIS Report).
All the empirical observations in physiology, biochemistry and genetics are telling us that the brain, like the rest of the human body, is not divided into what one might call 'soft' and 'hard' ware. Everything is 'wetware' in more senses than one - dynamic, flexible and fluid, up to and including genomic DNA. The brain needs experience to develop adequately, to make and break connections, to change gene function and structure. Experience not only creates memory, but also changes the brain and the body in its entirety and in unique ways.
A person is more like an ongoing narrative in which 'word' is literally made 'flesh'. She is created and recreated, molecule by molecule, cell by cell, as she goes about the business of living and experiencing. Every experience changes her whole being, and each encounter of apparently the same is not quite the same, even if one were to judge only by the firing patterns of nerve cells in the brain. 'Memories' are projections to the future.
As far as I can see, the downloaded 'brain in the computer' can, at best, be a frustrated psychotic without the flesh and blood body to live out her subjective experience. And the perfect, synthetic, 'nanoengineered' non-biodegradable body simply won't do. Why? Because all our memories - constituting a substantial part of our downloaded person - are those of the flesh and blood being, which may have no means of translating into the non-biodegradable perfect body.
The deeper, more important question is, can 'simulated life' - because that's all a computer can be - evolve into something indistinguishable form real life? Kurzweil offers little to convince me it can.
He starts by rewriting the history of creation from the big bang to the emergence of the human species, and beyond, to the computer; and it is the computer, not the human species, that's to be placed at the pinnacle of creation. So computer scientists - with a little help from nanotechnologists, genetic engineers and brain scientists - will take over where God left off.
This quartet is now officially constituted as NIBC (nano-info-bio-cogno-technologies), following a National Science Foundation workshop in the United States last December that generated a 450 page report. The agenda is to promote the combined role of the four technologies to accelerate "advancement of mental, physical and overall human performance", by means that include eugenics. There is no doubt that many of the participants have been inspired by Kurzweil (see "The Brave New World Quartet", this series).
Kurzweil draws analogy between 'Moore's law' - that the number of components on a chip as well as its speed doubles every two years - and the exponential speeding up of evolution. It took a long time for the first bacteria to emerge from the 'primordial soup', less time for multicellular organisms to come on the scene, then speeding up to vertebrates and homo sapiens, the creator of technology that also accelerates exponentially over time.
But won't Moore's law come up against a wall when the limit of miniaturisation is reached, and no more components can be packed on a chip? No, says Kurzweil, nanotechnologists are already developing molecular size electronics. Indeed, they are, and at the moment, it is already possible to pack about 20 times more components on a chip. And beyond that, quantum computing is on the horizon, which not only offers to solve problems that take infinitely long on an ordinary computer, but may solve others that are currently unsolvable as well.
Nanotechnology has its own strong advocates, the most prominent among them, Eric Drexler, whose 'visionary' ideas straddle science and science fiction (see box). Drexler too, has been inspired by neo-Darwinian genetic determinism. He predicts self-replicating nanoscale robots, raising fears they could take over the world, and molecular machines that could repair damaged cells, offering up the ultimate dream, again, of immortality.
The man who is said to have started nanotechnology is Eric Drexler, a space-systems researcher at MIT, who published Engines of Creation in 1986, in which he claimed that our ever improving ability to manipulate matter would lead to the creation of machine parts the size of small molecules. But Drexler says he was inspired by quantum physicist Richard Feynman, who gave an after-dinner talk in 1959 exploring the limits of miniaturization, and ended up by arguing the possibility, even the inevitability of "atom by atom" construction.
Drexler forsees the creation of machine parts the size of small molecules. These could be assembled into machines much smaller than biological cells, which could interact directly with the machinery of biology. Afterall, biology is nothing more than a bunch of molecular machines created and honed by evolution.
Drexler sees "molecular machines assemble molecular building blocks to form products, including new molecular machines." It is the ultimate dream of the computer scientist to realize the self-reproducing automata, or in this case, the self-reproducing nanrobot.
This alone caused much public alarm. So much so that Bill Joy, chief scientist of Sun Microsystems, wrote a long article in the Wired magazine in 2000 where he proposed that we should consider stopping the developments of nanotechnology for fear of being overrun by a mass of "grey goo" - self-replicating nanrobots.
Drexler continues, "Stepping beyond the biological analogy, it would be a natural goal to be able to put every atom in a selected place (where it would serve as part of some active or structural component) with no extra molecules on the loose to jam the works. Such a system would not be liquid or gas, as no molecules would move randomly, nor would it be a solid, in which molecules are fixed in place. Instead this new machine-phase matter would exhibit the molecular movements seen today only in liquids and gases as well as the mechanical strength typically associated with solids. Its volume would be filled with active machinery."
Drexler appears to be struggling towards the idea that the artificial molecular machines are like those of the living system, liquid crystalline with the texture of flesh (see main text). Living molecular machines are made up of at least twice their weight of 'biological' water - water that is an integral part of their structure and function. These molecular machines, densely packed and embedded in the liquid crystalline matrix runs in almost perfect cycles, drawing on coherent energy extracted and stored from metabolism. The living molecular machines basically borrow the coherent energy and return it only slightly degraded to the matrix. The efficiency of living molecular machines is such that they generate very little waste heat, which is why they can be packed so densely and work without burning out.
Of course, one of the biggest problems for artificial molecular machines, let alone self-replicating molecular machines, is the energy source. Another is energy dissipation - to get rid of the waste heat - neither of which Drexler has addressed. But there is a deeper problem.
The organism is run, in the ideal, on quantum coherence. And hence - this is the sting in the tail - the molecular machines cannot be individually controlled. Instead, the organism is a system of molecular democracy of distributed control. Each individual molecular machine operates with maximum freedom and is yet correlated with the whole.
What about "the most exciting goal" according to Drexler, "of the molecular repair of the human body"? Medical nanorobots that could destroy viruses and cancer cells, repair damaged structure, remove accumulated wastes from the brain and "bring the body back to a state of youthful health." Ah, the ultimate dream of immortality!
Unfortunately, our body's immune system may well see these artificial molecular machines as foreign invaders and try to get rid of them, or worse, they might clog up the immune system for good.
Have they all gone stark raving mad and megalomaniac? Thankfully, there's one lone voice of reason left from the IT revolutionaries.
Jaron Lanier, a pioneer in virtual reality, musician and currently the lead scientist for the National TeleImmersion Initiative, finally decides to take on Kurzweil, Drexler and others, whom he calls the "cybernetic totalists".
"There is a real chance that evolutionary psychology, artificial intelligence, Moore's Law fetishzing, and the rest of the package will catch on in a big way, as big as Freud or Marx did in their times. Or bigger, since these ideas might end up essentially built into the software that runs our society and our lives." Lanier warns, "If that happens, the ideology of cybernetic totalist intellectuals will be amplified from novelty into a force that could cause suffering for millions of people."
In other words, we could be misled into shaping our future according to this dreadful vision that will become a self-fulfilling prophecy.
Lanier refers in particular to the "eschatologies" of Kurzweil, Moravec and Drexler, which seem to follow directly and inevitably from an understanding of the world that has been most sharply articulated by biologist Richard Dawkins, and philosopher Daniel Dennett.
Richard Dawkins is responsible for the idea that evolution can be understood in terms of competition between 'selfish' genes, whose only mission is to replicate. The organism is merely the gene's way of begetting another gene. Ideas are said to spread in an analogous way as bit-ideas or 'memes'. Dennett subscribes wholeheartedly to this view, and sees humans as simply specialized computers.
So, cybernetic totalists look at culture and see 'memes' that compete for brain space in humans, rather like viruses. And once we have reduced ideas into meaningless bits, then "any particular reshuffling of its bits seem unimportant". "We kid ourselves when we think we understand something, even a computer, merely because we can model or digitise it." Lanier says.
There's a plethora of recent theories in which the brain is said to produce random distribution of subconscious ideas that compete with one another, Darwinian fashion, until only the best has survived. But do these theories really fit with what people do?
Over a decade of work world wide in Darwinian approaches to generating software has produced nothing that would make software in general any better, Lanier points out.
People have confused ideal computers with real computers. Although 'self-reproducing automata' are possible in theory, someone's got to write the software that gets the process going, and humans have given absolutely no evidence of being able to write such software.
Just because the computer can get faster doesn't mean it gets smarter. Moore's law in hardware development must be starkly contrasted with "the Great Shame of computer science," says Lanier, "which is that we don't seem to be able to write software much better as computers get much faster".
What worries Lanier, is decidedly not that we shall be enslaved by computers. His own vision of terror is this. The biotech industry is setting itself up for decades of expensive software trouble. While there are all sorts of useful databases and modelling packages developed by biotech firms and labs, they all exist in isolated bubbles. So vast resources will be expended to get data from one bubble to another.
"If Moore's Law is upheld for another 20 to 30 years, there will not only be a vast amount of computation going on planet earth but also the maintenance of that computation will consume the efforts of almost every living person." We may end up with a planetful of help desks!
I have argued that's precisely why it is time to abandon the mausoleums of useless genomic information and employ scientists in more imaginative, less mind-numbing research.
The solution is to keep all genomic data on the public Genbank, where they can be accessed freely by everyone, everywhere.
Lanier concludes: "Treating technology as if it were autonomous is the ultimate self-fulfilling prophecy. There is no difference between machine autonomy and the abdication of human responsibility."
Article first published 22/08/02
Got something to say about this page? Comment