Phew... I was about to buy a chain to lock down my computer.
Computers won’t outsmart us any time soon
Published at 12:01AM, August 8 2014
Fear of artificial intelligence is unfounded — we do not understand human consciousness so we can hardly create it
The anguish of Frankenstein’s creature is terribly recognisable. Offended by his creator’s rejection, the literate and articulate monster pleads for a girlfriend but Dr Frankenstein stops work when he realises that he might be creating a breed of demons. Stricken by grief, the monster kills Frankenstein’s wife in revenge. Mary Shelley’s monster has an emotional register that runs through hope, desire and regret. He is like us, only more so.
The fear is raising its oversized head again. Last month researchers at the Royal Society claimed that a computer had passed the Turing test and thereby achieved human intelligence. The inventor Elon Musk declared artificial intelligence to be a greater risk to the future of humanity than nuclear war, an idea borrowed from Superintelligence, by Niklas Boström, a professor of philosophy at Oxford.
Boström argues that as a result of a minor programming error superintelligent beings will develop their own malign purposes and requisition the resources that humans need to live on Earth. It is the story of 2001: A Space Odyssey, The Terminator , Robocop and, with a kinder ethical twist, the writings of Isaac Asimov and Iain M Banks. Stephen Hawking has lent his authority to this perennial fear which the advance of technology seems to generate. But it is, to use a technical term from the philosophy of mind, complete rubbish.
To start with, the Turing test is a fraudulent way to judge intelligence. The idea is that if a computer’s response, via email or telex, to questions fools a third of the assembled judges into believing it is human then that computer should be credited with human-level consciousness. On June 7 a Russian chatterbot given the identity of a 13-year-old Ukrainian boy and the name of Eugene Goostman, convinced a third of the judges, including (I kid you not) an actor from Red Dwarf, that, though its grasp of English grammar was tenuous, it was human.
The obvious objection is that a computer isn’t conscious just because it fools a bloke from Red Dwarf. This was put in a more systematic way in 1980 by John Searle who imagined a man sitting in a closed room. Every now and then a piece of paper covered in squiggles is passed to him, whereupon he consults a manual and transcribes other squiggles according to its instructions. Quite unknown to the man in the room, all his meaningless marks are Chinese script. He will pass the Turing test but it would be absurd to say that he is conscious of what he is doing, namely writing Chinese.
It is for this reason that malign computers really will be consigned to the pages of science fiction for a long while yet. Progress in computing capacity has been truly astonishing. Genius machines can do complex mathematics, medical diagnosis, select stocks for a mutual fund portfolio and beat Garry Kasparov at chess. Consciousness, though, is something greater than facts and figures. There is no sense of a design for life in the reliable processing of a machine. For all its ability with numbers, the supercomputer is, as Steven Pinker points out in How The Mind Works, about equal to the nervous system of a snail. They struggle to mimic the intuitive creativity that comes naturally to human beings. A computer has no trouble remembering a 25 digit number but will struggle to recall the basic gist of Little Red Riding Hood.
The best way to think about this is to ask, in the title of a famous essay by Thomas Nagel, What is it like to be a bat? The answer is that we cannot ever know but it is definitely like something. If I were to sleep upside down and move through the city streets by sonic echo-location, then I might get some understanding of what it is like for me to be a bat (more likely I’d just make a fool of myself). But I would not get any closer to understanding what it is like for a bat to be a bat.
That is Nagel’s point. There is something about being a bat that can only be seen from the inside. There is a mindset of being a bat but there is no mindset of being a machine. I can try to convey in words the experience of being me, but nobody else can truly know what it is like. Consciousness is like Louis Armstrong’s reply when he was asked to define jazz: “Lady, if you gotta ask, you don’t know.”
The question then arises of what this experience is made of. There are three general approaches. The first was begun by René Descartes, who wrote that the physical brain and the mental mind were made of separate stuff. Descartes thought that the join could be found in the pineal gland in the brain. Nobody since, however, has managed to explain how the mind can change the physical world without violating the laws of physics.
The separation of mind and body was then replaced by a physical account in which consciousness was said to be identical with the flow of neurological fluids in the brain. Quite how the peculiar experience of what it is like to be me is derived from raw chemicals is not clear. For all the knowledge disclosed by positron emission topography and magnetic resonance imaging, scientists do not begin to understand how the firing synapses in the brain create the wonder of experience.
This leads to the third position, which is that consciousness passes all understanding. Perhaps the people around me are zombies who merely walk and talk like humans? (Trust me, where I work, that’s a real possibility). It may be, as Nagel and Pinker have suggested, that we are biologically incapable of working this one out. “If so,” writes Pinker, “our invention the computer would present us with the ultimate tease.”
There is no reason to expect that a computer can ever be more than a complex box in which unconscious electrical impulses pass through lifeless circuits. The computer can no more said to be thinking than a clock can be said to be telling the time. The monster of Dr Frankenstein had independent designs, not least on women. He was what we always imagine artificial intelligence to be — like us, only more so.
This is a science fiction that we can lay aside. The Liberians with the ebola virus are getting a gruesome answer to the question of what it is like to be a bat, but it is habitually the purposes to which human beings set their minds that we should fear. The creations of conscious minds, religion and ideology are out there killing while the monsters are stuck at home either moaning about not having a girlfriend or not knowing what a girlfriend is in the first place.
http://www.thetimes.co.uk/tto/opinion/c ... 169817.ece
