Could a machine ever be conscious? Many experts say ‘yes, of course, believing that technology will one day be able to produce a machine with a mind of its own. But these experts have been proved wrong in the past – and according to earlier predictions, conscious machines should exist by now.
If you simulated a weather pattern on a computer, would it be real weather? Would its wind ruffle your hair, or does its rain make you wet? No, of course not, no matter how exact and complete the simulation. So then, why would you expect a simulation of your brain processes to be really conscious? On the other hand, think about this. Suppose just one of your brain cells was replaced with a microchip that did the same job. If all the inputs and outputs remained precisely the same, you would never notice the difference. And what if another, then another, cell was replaced, until gradually all 100 billion neurons in your brain had been turned to silicon? Would there ever come to a point where you stopped being you? Would you become, in effect, a conscious machine?
These are exactly the kinds of questions that cognitive scientists have been throwing at each other since the dawn of the computer era. Even now, science does not seem much closer to answering them.
Applying Some Logic
Alan Turing, the British mathematician who laid the foundations of! modern computing in the 1940s firmly believed that machines would be able to think for themselves sooner rather than later, and certainly by the end of the 20th century. Turing’s thorough was to prove that, in theory, the simplest computing device – he imagined a single key making or erasing marks in succession on an infinite tape – could carry out any logically defined operation. All computers were, therefore, identical under the skin, as their workings could always be described in terms of this theoretical device (which became known as a Turing machine). You could build a computer out of anything, even string and old tin cans. While it might not be very efficient or fast, the output of its programs would remain the same. Turing offered mathematical proof that if you could describe the ‘programs’ being run on the human brain, then the same programs could be run on an artificial brain. He did not think that the technology of his day would have to be scaled up much to run the basic programs of human consciousness at useful speeds.
There was, as Turing himself admitted, a potential problem with this argument, which lay in whether the brain’s operations were indeed logical. Could whatever brains did actually be represented as a binary sequence of 0s and 1s? But like most scientists, Turing was optimistic, because brain cells seemed to communicate using on-off patterns of electrical spikes.
So, there was a clever intellectual trick at the heart of Turing’s argument. He had proved that, in mathematical terms, all computers were essentially the same. Now, it was up to others to prove that human brains were not computers. This was, and remains, a very difficult task. Anyone who draws a categorical distinction between the logical workings of a brain and those of a computer can always be undermined – they simply have not considered every last alternative or explored every fact. While you can show that machine consciousness is highly unlikely or impractical, scientifically speaking, you can not remove the doubt about whether it might be done in the future based on some as-yet-undiscovered fact.
Chinese whispers
Philosophers, of course, have made valiant attempts to come up with just such an ‘in principle’ argument against machine consciousness. The most famous of these is the Chinese Room thought experiment of US philosopher John Searle.
Searle asks us to imagine a man locked in a room. He does not understand Chinese, but he is surrounded by millions of books that give the precise answer to every possible question framed in that language. From time to time, someone shoves a query written in Chinese through a slot. The man’s job is to rifle through the books, find a march to the question, and then scribble back the prescribed reply. From the outside, the room appears to have an intelligent comprehension of Chinese. And yet we know that inside there is just a man following rules without having the slightest clue about the meaning of either the question or the response. Searle says that the same would be the case with a computer. It might run a program. and output realistic behavior. But there would be no inner light of experience or understanding. A computer has rules, but not semantics – that is, inner knowledge of what in the real world is being manipulated by those rules.
But computer scientists disagree with this viewpoint. They say that our brains have semantics and true understanding because they manipulate mental representations – broad washes of sensation and memory displayed across millions of brain cells. So, the correct way to imagine the Chinese Room is as many little men, each representing points of data in a coordinated show. Like individual brain cells, each time, the figure would merely follow local processing rules. But the network as a whole would be conscious. As a system, the Chinese Room really would ‘feel’ that it understood the questions.
Again, the argument is one that the anti-computer camp seemingly cannot win. Searle has continued to blaze away, objecting that simulated weather will never make you wet and a simulated carburetor will never power a car. Simulations can not have real effects on the world. But computer scientists reply that a weather model hooked up to your garden sprinkler would certainly make you wet. And in the same way, an artificial brain equipped with eyes to see and hands to act could be just as much a part of the world as you.
On the basis of this line of argument, at least, there seems no good reason to rule out the possibility of machine consciousness in principle. On the other hand, 50 years of intensive but largely fruitless research has shown that Turing and many others were wildly optimistic in their forecasts for machine intelligence, let alone machine consciousness.
The unrealized dream
During the 1970s and 1980s, huge amounts of money were plowed into artificial intelligence (Al ) research by governments, industry, and especially defense research agencies. Japan, Europe, and the US became locked into an intellectual arms race. The reasoning was that if the technocrats failed to deliver the conscious machines they were promising, then even slightly smart ones would have valuable applications. History tells us that the AI movement wa. Largely unproductive. It did give us one or two clever new programming tricks, but the commercial computers of today are much the same as those of yesteryear – the main difference is that they are smaller, cheaper, and faster.
There are still die-hards who insist that a breakthrough is waiting around the corner. They point to the promise of neural networks- computers designed in direct imitation of brain networks. These may have the equivalent of only a few thousand brain cells at present (about enough to power a cockroach), but it might be just 30 or 40 years before such a computer has enough connections to equal a human brain. However, even most computer scientists have grown wary of sweeping pronouncements, and some feel that any estimates should be made in terms of decades or centuries rather than years.
The dream of conscious machines will never disappear because science has no way of proving it impossible. Besides, people seem to be too enamored with the idea to give it up. However, there is not much for this being placed on it happening in the near future.
Again, the argument is one that the. anti-computer camp seemingly cannot win. Searle has continued to blaze away, objecting that simulated weather will never make you wet and a simulated carburetor will never power a car. Simulations cannot have real effects on the world. But computer scientists reply that a leather model hooked up to your garden sprinkler would certainly make you wet. And in the same way, an artificial brain equipped with eyes to see and hands to act could be just as much a part of the world as you.
On the basis of this line of argument, at least, there seems no good reason to rule out the possibility of machine consciousness in principle. On the other hand, 50 years of intensive but largely fruitless research has shown that Turing and many others were wildly optimistic in their forecasts for machine intelligence, let alone machine consciousness.
The unrealized dream
During the 1970s and 1980s, huge amounts of money were plowed into artificial intelligence (AI) research by governments, industry, and especially defense research agencies. Ja pan, E u rope, and the US became locked into an intellectual arms race. The reasoning was that if the technocrats failed to deliver the conscious machines they were promising, then even slightly smart ones would have valuable applications. History tells us that the AT movement was largely unproductive. It did give us one or n;vo clever new progra1m Tting tricks, but the commercial comp ut ers of today are much the same as those of yesteryear – the main difference is that they are smaller, cheaper, and faster.
There are still die-hards who insist that a breakthrough is waiting around the corner. They point to the promise of neural networks – computers designed in direct i1rutation of brain networks. These may have the equivalent of only a few thousand brain cells at present (about enough to power a cockroach), but it might be just 30 or 40 years before such a computer has enough connections to equal a human brain. However, even most computer scientists have grown wary of sweeping pronouncements, and some feel that any estimates should be made in terms of decades or centuries rather than years.
The dream of conscious machines will never disappear because science has no way of proving it impossible. Besides, people seem to be too enamored with the idea to give it up. However, not much faith is being placed in it happening in the near future.
A Conscious Internet
What if the internet one day grew so connected that it woke up and became conscious? Some experts think this could really happen. Belgian computer scientist Francis Heylighen argues that web pages are like information-containing brain cells, and the hyperlinks between them are like synaptic connections between brain cells. Throw in search engines and other kinds of intelligent programs, and perhaps the whole thing could start to come alive.
It’s an interesting idea. However, the more sober-minded point out that machines are best seen as amplifiers of human activity. The Industrial Revolution was about the amplification of human muscle power. Today’s information revolution is about the amplification of the human mind’s power. So while the internet does promise remarkable things, the real story will not be its emerging consciousness, but how it will extend the reach of our own minds into a global, shareable body of knowledge and culture.
Stranger than fiction
The science of artificial intelligence is never far from science fiction. In the imaginations of writers and filmmakers, robots like Star Wars’ C3PO are given human characteristics such as reflection, self-awareness, and self-interest.
